diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,15016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.6972165648336728,
+  "global_step": 50000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999999431403446e-05,
+      "loss": 1.8514,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999977256144297e-05,
+      "loss": 1.2222,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999948826348907e-05,
+      "loss": 1.1193,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999909024680632e-05,
+      "loss": 1.0893,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999857851184724e-05,
+      "loss": 1.0828,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999795305919377e-05,
+      "loss": 1.0137,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999721388955722e-05,
+      "loss": 1.0248,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999636100377818e-05,
+      "loss": 1.0394,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999539440282648e-05,
+      "loss": 1.0873,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999943140878014e-05,
+      "loss": 1.1132,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999931200599314e-05,
+      "loss": 1.0493,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999181232057437e-05,
+      "loss": 1.0894,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999039087121746e-05,
+      "loss": 1.1671,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998885571347712e-05,
+      "loss": 1.0525,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998720684909913e-05,
+      "loss": 1.0033,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999854442799585e-05,
+      "loss": 1.0375,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998356800805973e-05,
+      "loss": 1.0725,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998157803553638e-05,
+      "loss": 1.0104,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997947436465157e-05,
+      "loss": 1.0205,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997725699779747e-05,
+      "loss": 1.079,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999749259374957e-05,
+      "loss": 1.0023,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999724811863971e-05,
+      "loss": 1.0257,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999699227472818e-05,
+      "loss": 0.9901,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996725062305934e-05,
+      "loss": 0.9934,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999644648167684e-05,
+      "loss": 1.056,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996156533157692e-05,
+      "loss": 1.0688,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999585521707822e-05,
+      "loss": 1.1045,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999554253378108e-05,
+      "loss": 1.0399,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995218483621858e-05,
+      "loss": 1.0219,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994883066969055e-05,
+      "loss": 1.0287,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994536284204106e-05,
+      "loss": 1.0638,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994178135721372e-05,
+      "loss": 1.0247,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999380862192813e-05,
+      "loss": 0.99,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99934277432446e-05,
+      "loss": 1.0635,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993035500103904e-05,
+      "loss": 0.9725,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992631892952108e-05,
+      "loss": 1.0221,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999221692224818e-05,
+      "loss": 1.0333,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991790588464034e-05,
+      "loss": 1.0619,
+      "step": 760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9991352892084483e-05,
+      "loss": 0.9604,
+      "step": 780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999090383360728e-05,
+      "loss": 0.9817,
+      "step": 800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9990443413543086e-05,
+      "loss": 1.0362,
+      "step": 820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998997163241549e-05,
+      "loss": 1.0436,
+      "step": 840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9989488490761e-05,
+      "loss": 1.0088,
+      "step": 860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988993989129035e-05,
+      "loss": 1.06,
+      "step": 880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988488128081943e-05,
+      "loss": 1.0384,
+      "step": 900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998797090819499e-05,
+      "loss": 0.9912,
+      "step": 920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9987442330056352e-05,
+      "loss": 1.0263,
+      "step": 940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9986902394267117e-05,
+      "loss": 0.9909,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9986351101441308e-05,
+      "loss": 0.993,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9985788452205847e-05,
+      "loss": 1.0823,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9985214447200566e-05,
+      "loss": 0.9748,
+      "step": 1020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9984629087078234e-05,
+      "loss": 0.9944,
+      "step": 1040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998403237250451e-05,
+      "loss": 1.0813,
+      "step": 1060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9983424304157976e-05,
+      "loss": 0.9385,
+      "step": 1080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998280488273012e-05,
+      "loss": 1.0154,
+      "step": 1100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9982174108925346e-05,
+      "loss": 0.9694,
+      "step": 1120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998153198346097e-05,
+      "loss": 0.9958,
+      "step": 1140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9980878507067207e-05,
+      "loss": 1.1112,
+      "step": 1160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9980213680487186e-05,
+      "loss": 1.025,
+      "step": 1180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9979537504476945e-05,
+      "loss": 1.025,
+      "step": 1200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.997884997980543e-05,
+      "loss": 0.9974,
+      "step": 1220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9978151107254477e-05,
+      "loss": 1.0615,
+      "step": 1240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9977440887618855e-05,
+      "loss": 0.9963,
+      "step": 1260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9976719321706213e-05,
+      "loss": 0.9929,
+      "step": 1280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.997598641033711e-05,
+      "loss": 1.0235,
+      "step": 1300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.997524215434501e-05,
+      "loss": 0.9834,
+      "step": 1320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974486554576274e-05,
+      "loss": 0.9981,
+      "step": 1340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973719611890164e-05,
+      "loss": 0.914,
+      "step": 1360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972941327158848e-05,
+      "loss": 0.9992,
+      "step": 1380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997215170126738e-05,
+      "loss": 0.9329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971350735113718e-05,
+      "loss": 1.0179,
+      "step": 1420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970538429608716e-05,
+      "loss": 1.0618,
+      "step": 1440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996971478567612e-05,
+      "loss": 1.02,
+      "step": 1460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996887980425258e-05,
+      "loss": 1.0123,
+      "step": 1480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996803348628762e-05,
+      "loss": 1.0023,
+      "step": 1500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967175832743677e-05,
+      "loss": 0.9477,
+      "step": 1520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9966306844596063e-05,
+      "loss": 0.9968,
+      "step": 1540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9965426522832985e-05,
+      "loss": 0.9729,
+      "step": 1560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9964534868455544e-05,
+      "loss": 1.0453,
+      "step": 1580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9963631882477716e-05,
+      "loss": 1.0657,
+      "step": 1600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9962717565926376e-05,
+      "loss": 0.9159,
+      "step": 1620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961791919841275e-05,
+      "loss": 0.9438,
+      "step": 1640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960854945275055e-05,
+      "loss": 0.9773,
+      "step": 1660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995990664329323e-05,
+      "loss": 1.0714,
+      "step": 1680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958947014974216e-05,
+      "loss": 0.9914,
+      "step": 1700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957976061409277e-05,
+      "loss": 0.9779,
+      "step": 1720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956993783702594e-05,
+      "loss": 0.9906,
+      "step": 1740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956000182971192e-05,
+      "loss": 0.9829,
+      "step": 1760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954995260344993e-05,
+      "loss": 1.0036,
+      "step": 1780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995397901696679e-05,
+      "loss": 0.9597,
+      "step": 1800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995295145399224e-05,
+      "loss": 0.967,
+      "step": 1820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951912572589892e-05,
+      "loss": 1.0017,
+      "step": 1840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950862373941145e-05,
+      "loss": 0.9761,
+      "step": 1860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994980085924028e-05,
+      "loss": 0.9761,
+      "step": 1880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994872802969445e-05,
+      "loss": 1.0116,
+      "step": 1900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.994764388652366e-05,
+      "loss": 1.0121,
+      "step": 1920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9946548430960796e-05,
+      "loss": 0.9758,
+      "step": 1940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9945441664251605e-05,
+      "loss": 0.9948,
+      "step": 1960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9944323587654685e-05,
+      "loss": 1.0505,
+      "step": 1980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9943194202441514e-05,
+      "loss": 0.9806,
+      "step": 2000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9942053509896417e-05,
+      "loss": 0.9377,
+      "step": 2020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9940901511316585e-05,
+      "loss": 1.0366,
+      "step": 2040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993973820801206e-05,
+      "loss": 0.9959,
+      "step": 2060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993856360130574e-05,
+      "loss": 1.037,
+      "step": 2080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9937377692533382e-05,
+      "loss": 1.0662,
+      "step": 2100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99361804830436e-05,
+      "loss": 1.0312,
+      "step": 2120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934971974197843e-05,
+      "loss": 0.9109,
+      "step": 2140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993375216737042e-05,
+      "loss": 0.9742,
+      "step": 2160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932521063948494e-05,
+      "loss": 1.0021,
+      "step": 2180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931278665332057e-05,
+      "loss": 0.9514,
+      "step": 2200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9930024972933966e-05,
+      "loss": 0.9497,
+      "step": 2220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9928759988179908e-05,
+      "loss": 0.9337,
+      "step": 2240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.992748371250841e-05,
+      "loss": 0.9945,
+      "step": 2260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.992619614737085e-05,
+      "loss": 0.9873,
+      "step": 2280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9924897294231434e-05,
+      "loss": 0.9715,
+      "step": 2300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9923587154567214e-05,
+      "loss": 0.946,
+      "step": 2320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9922265729868065e-05,
+      "loss": 0.9688,
+      "step": 2340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.992093302163671e-05,
+      "loss": 0.9837,
+      "step": 2360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9919589031388688e-05,
+      "loss": 1.0015,
+      "step": 2380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.991823376065238e-05,
+      "loss": 0.9847,
+      "step": 2400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.991686721096899e-05,
+      "loss": 0.9702,
+      "step": 2420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.991548938389255e-05,
+      "loss": 0.9749,
+      "step": 2440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.991410028098991e-05,
+      "loss": 0.968,
+      "step": 2460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9912699903840754e-05,
+      "loss": 0.9446,
+      "step": 2480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.991128825403758e-05,
+      "loss": 0.9667,
+      "step": 2500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9909865333185704e-05,
+      "loss": 0.9359,
+      "step": 2520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9908431142903263e-05,
+      "loss": 1.0054,
+      "step": 2540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9906985684821208e-05,
+      "loss": 0.9583,
+      "step": 2560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9905528960583305e-05,
+      "loss": 0.9364,
+      "step": 2580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990406097184613e-05,
+      "loss": 0.9839,
+      "step": 2600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9902581720279074e-05,
+      "loss": 1.0165,
+      "step": 2620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9901091207564326e-05,
+      "loss": 0.9603,
+      "step": 2640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9899589435396888e-05,
+      "loss": 0.9977,
+      "step": 2660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9898076405484565e-05,
+      "loss": 0.9618,
+      "step": 2680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9896552119547964e-05,
+      "loss": 0.89,
+      "step": 2700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9895016579320495e-05,
+      "loss": 1.0115,
+      "step": 2720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.989346978654836e-05,
+      "loss": 0.8936,
+      "step": 2740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9891911742990567e-05,
+      "loss": 0.9733,
+      "step": 2760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9890342450418907e-05,
+      "loss": 0.9443,
+      "step": 2780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9888761910617967e-05,
+      "loss": 1.016,
+      "step": 2800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9887170125385135e-05,
+      "loss": 0.8977,
+      "step": 2820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9885567096530566e-05,
+      "loss": 0.9582,
+      "step": 2840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9883952825877225e-05,
+      "loss": 1.0007,
+      "step": 2860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.988232731526084e-05,
+      "loss": 1.0113,
+      "step": 2880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9880690566529933e-05,
+      "loss": 0.9572,
+      "step": 2900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.987904258154581e-05,
+      "loss": 1.0198,
+      "step": 2920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.987738336218254e-05,
+      "loss": 0.8695,
+      "step": 2940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9875712910326982e-05,
+      "loss": 0.9847,
+      "step": 2960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.987403122787876e-05,
+      "loss": 0.9547,
+      "step": 2980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9872338316750264e-05,
+      "loss": 0.9653,
+      "step": 3000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9870634178866676e-05,
+      "loss": 0.957,
+      "step": 3020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.986891881616592e-05,
+      "loss": 0.976,
+      "step": 3040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9867192230598692e-05,
+      "loss": 0.9866,
+      "step": 3060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.986545442412846e-05,
+      "loss": 1.0012,
+      "step": 3080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9863705398731444e-05,
+      "loss": 0.9578,
+      "step": 3100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.986194515639662e-05,
+      "loss": 0.9816,
+      "step": 3120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.986017369912573e-05,
+      "loss": 0.9501,
+      "step": 3140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9858391028933258e-05,
+      "loss": 0.9618,
+      "step": 3160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9856597147846447e-05,
+      "loss": 0.9178,
+      "step": 3180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.985479205790528e-05,
+      "loss": 0.9475,
+      "step": 3200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.98529757611625e-05,
+      "loss": 0.9608,
+      "step": 3220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9851148259683586e-05,
+      "loss": 0.8593,
+      "step": 3240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.984930955554676e-05,
+      "loss": 0.9459,
+      "step": 3260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.984745965084298e-05,
+      "loss": 0.8713,
+      "step": 3280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9845598547675945e-05,
+      "loss": 0.9075,
+      "step": 3300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9843726248162096e-05,
+      "loss": 0.9818,
+      "step": 3320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.984184275443059e-05,
+      "loss": 0.9602,
+      "step": 3340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.983994806862333e-05,
+      "loss": 0.9412,
+      "step": 3360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9838042192894935e-05,
+      "loss": 0.9837,
+      "step": 3380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9836125129412756e-05,
+      "loss": 1.0087,
+      "step": 3400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9834196880356863e-05,
+      "loss": 1.0303,
+      "step": 3420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.983225744792005e-05,
+      "loss": 0.9416,
+      "step": 3440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9830306834307828e-05,
+      "loss": 0.9829,
+      "step": 3460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9828345041738413e-05,
+      "loss": 0.9906,
+      "step": 3480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9826372072442747e-05,
+      "loss": 0.9868,
+      "step": 3500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9824387928664478e-05,
+      "loss": 0.9515,
+      "step": 3520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9822392612659964e-05,
+      "loss": 0.9713,
+      "step": 3540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9820386126698256e-05,
+      "loss": 0.8813,
+      "step": 3560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.981836847306112e-05,
+      "loss": 0.9842,
+      "step": 3580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.981633965404302e-05,
+      "loss": 0.9981,
+      "step": 3600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9814299671951113e-05,
+      "loss": 0.9399,
+      "step": 3620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9812248529105254e-05,
+      "loss": 0.8522,
+      "step": 3640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9810186227837986e-05,
+      "loss": 0.9817,
+      "step": 3660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9808112770494546e-05,
+      "loss": 0.9788,
+      "step": 3680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9806028159432854e-05,
+      "loss": 0.9442,
+      "step": 3700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9803932397023512e-05,
+      "loss": 1.0024,
+      "step": 3720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9801825485649816e-05,
+      "loss": 1.0184,
+      "step": 3740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9799707427707723e-05,
+      "loss": 0.99,
+      "step": 3760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9797578225605875e-05,
+      "loss": 1.0291,
+      "step": 3780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.979543788176559e-05,
+      "loss": 0.9496,
+      "step": 3800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.979328639862085e-05,
+      "loss": 0.9149,
+      "step": 3820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9791123778618303e-05,
+      "loss": 0.9756,
+      "step": 3840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.978895002421727e-05,
+      "loss": 0.9599,
+      "step": 3860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.978676513788973e-05,
+      "loss": 0.9328,
+      "step": 3880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.978456912212032e-05,
+      "loss": 0.9399,
+      "step": 3900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9782361979406334e-05,
+      "loss": 0.9641,
+      "step": 3920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9780143712257715e-05,
+      "loss": 0.9505,
+      "step": 3940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9777914323197063e-05,
+      "loss": 1.0097,
+      "step": 3960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.977567381475963e-05,
+      "loss": 0.9627,
+      "step": 3980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.97734221894933e-05,
+      "loss": 0.9573,
+      "step": 4000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9771159449958612e-05,
+      "loss": 0.974,
+      "step": 4020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9768885598728734e-05,
+      "loss": 0.9207,
+      "step": 4040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.976660063838947e-05,
+      "loss": 0.9765,
+      "step": 4060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9764304571539266e-05,
+      "loss": 0.9475,
+      "step": 4080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9761997400789194e-05,
+      "loss": 0.9764,
+      "step": 4100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9759679128762946e-05,
+      "loss": 0.9277,
+      "step": 4120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9757349758096855e-05,
+      "loss": 0.9738,
+      "step": 4140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.975500929143986e-05,
+      "loss": 0.9991,
+      "step": 4160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9752657731453523e-05,
+      "loss": 0.9421,
+      "step": 4180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9750295080812025e-05,
+      "loss": 0.972,
+      "step": 4200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.974792134220215e-05,
+      "loss": 0.967,
+      "step": 4220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9745536518323304e-05,
+      "loss": 0.9693,
+      "step": 4240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9743140611887487e-05,
+      "loss": 0.9242,
+      "step": 4260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.974073362561931e-05,
+      "loss": 1.0251,
+      "step": 4280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9738315562255983e-05,
+      "loss": 1.0025,
+      "step": 4300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9735886424547306e-05,
+      "loss": 1.0041,
+      "step": 4320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.973344621525568e-05,
+      "loss": 0.8858,
+      "step": 4340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.97309949371561e-05,
+      "loss": 1.0171,
+      "step": 4360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9728532593036132e-05,
+      "loss": 0.9639,
+      "step": 4380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.972605918569594e-05,
+      "loss": 0.9343,
+      "step": 4400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.972357471794827e-05,
+      "loss": 0.8996,
+      "step": 4420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.972107919261844e-05,
+      "loss": 0.9477,
+      "step": 4440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9718572612544345e-05,
+      "loss": 1.0198,
+      "step": 4460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9716054980576446e-05,
+      "loss": 0.8851,
+      "step": 4480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.971352629957778e-05,
+      "loss": 0.9439,
+      "step": 4500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9710986572423947e-05,
+      "loss": 0.9632,
+      "step": 4520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9708435802003105e-05,
+      "loss": 1.0028,
+      "step": 4540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9705873991215973e-05,
+      "loss": 0.9911,
+      "step": 4560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.970330114297583e-05,
+      "loss": 0.9854,
+      "step": 4580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.970071726020849e-05,
+      "loss": 0.964,
+      "step": 4600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.969812234585233e-05,
+      "loss": 0.9546,
+      "step": 4620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9695516402858277e-05,
+      "loss": 0.8992,
+      "step": 4640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9692899434189786e-05,
+      "loss": 0.9148,
+      "step": 4660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.969027144282285e-05,
+      "loss": 0.9543,
+      "step": 4680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9687632431746014e-05,
+      "loss": 1.0007,
+      "step": 4700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9684982403960332e-05,
+      "loss": 0.9029,
+      "step": 4720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.968232136247941e-05,
+      "loss": 0.9436,
+      "step": 4740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9679649310329353e-05,
+      "loss": 0.9336,
+      "step": 4760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9676966250548807e-05,
+      "loss": 0.9395,
+      "step": 4780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.967427218618893e-05,
+      "loss": 0.9229,
+      "step": 4800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.967156712031339e-05,
+      "loss": 0.9191,
+      "step": 4820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.966885105599837e-05,
+      "loss": 0.913,
+      "step": 4840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9666123996332564e-05,
+      "loss": 0.9655,
+      "step": 4860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.966338594441716e-05,
+      "loss": 1.0067,
+      "step": 4880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9660636903365854e-05,
+      "loss": 0.9048,
+      "step": 4900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9657876876304835e-05,
+      "loss": 0.9761,
+      "step": 4920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.965510586637279e-05,
+      "loss": 0.8365,
+      "step": 4940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.965232387672089e-05,
+      "loss": 0.9765,
+      "step": 4960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9649530910512796e-05,
+      "loss": 0.864,
+      "step": 4980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9646726970924645e-05,
+      "loss": 0.9732,
+      "step": 5000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9643912061145065e-05,
+      "loss": 0.8954,
+      "step": 5020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9641086184375148e-05,
+      "loss": 0.956,
+      "step": 5040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9638249343828463e-05,
+      "loss": 0.9609,
+      "step": 5060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9635401542731043e-05,
+      "loss": 1.0167,
+      "step": 5080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.963254278432139e-05,
+      "loss": 0.9232,
+      "step": 5100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9629673071850463e-05,
+      "loss": 0.9191,
+      "step": 5120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.962679240858168e-05,
+      "loss": 0.8266,
+      "step": 5140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9623900797790914e-05,
+      "loss": 0.9257,
+      "step": 5160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9620998242766482e-05,
+      "loss": 0.8912,
+      "step": 5180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9618084746809147e-05,
+      "loss": 0.8733,
+      "step": 5200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9615160313232122e-05,
+      "loss": 0.9486,
+      "step": 5220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.961222494536105e-05,
+      "loss": 0.9723,
+      "step": 5240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.960927864653401e-05,
+      "loss": 1.0091,
+      "step": 5260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9606321420101514e-05,
+      "loss": 0.9226,
+      "step": 5280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.96033532694265e-05,
+      "loss": 0.9532,
+      "step": 5300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.960037419788433e-05,
+      "loss": 0.9176,
+      "step": 5320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9597384208862776e-05,
+      "loss": 0.9545,
+      "step": 5340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9594383305762045e-05,
+      "loss": 0.9162,
+      "step": 5360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9591371491994735e-05,
+      "loss": 0.9597,
+      "step": 5380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.958834877098586e-05,
+      "loss": 0.9209,
+      "step": 5400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9585315146172843e-05,
+      "loss": 0.9597,
+      "step": 5420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9582270621005494e-05,
+      "loss": 0.9402,
+      "step": 5440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9579215198946028e-05,
+      "loss": 0.8907,
+      "step": 5460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9576148883469054e-05,
+      "loss": 0.9002,
+      "step": 5480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9573071678061565e-05,
+      "loss": 0.9352,
+      "step": 5500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.956998358622293e-05,
+      "loss": 0.9039,
+      "step": 5520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9566884611464916e-05,
+      "loss": 0.926,
+      "step": 5540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9563774757311646e-05,
+      "loss": 0.9952,
+      "step": 5560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.956065402729963e-05,
+      "loss": 0.8673,
+      "step": 5580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9557522424977745e-05,
+      "loss": 0.9081,
+      "step": 5600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9554379953907217e-05,
+      "loss": 0.9287,
+      "step": 5620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.955122661766165e-05,
+      "loss": 0.9262,
+      "step": 5640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.954806241982699e-05,
+      "loss": 0.937,
+      "step": 5660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.954488736400155e-05,
+      "loss": 0.8884,
+      "step": 5680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9541701453795976e-05,
+      "loss": 0.9416,
+      "step": 5700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9538504692833263e-05,
+      "loss": 0.9081,
+      "step": 5720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9535297084748747e-05,
+      "loss": 0.8875,
+      "step": 5740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9532078633190097e-05,
+      "loss": 0.9085,
+      "step": 5760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9528849341817316e-05,
+      "loss": 0.9679,
+      "step": 5780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9525609214302727e-05,
+      "loss": 0.9845,
+      "step": 5800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9522358254330987e-05,
+      "loss": 0.9942,
+      "step": 5820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.951909646559906e-05,
+      "loss": 0.9219,
+      "step": 5840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.951582385181623e-05,
+      "loss": 0.9227,
+      "step": 5860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9512540416704095e-05,
+      "loss": 0.9447,
+      "step": 5880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.950924616399655e-05,
+      "loss": 0.9101,
+      "step": 5900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9505941097439805e-05,
+      "loss": 0.9938,
+      "step": 5920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9502625220792345e-05,
+      "loss": 0.9749,
+      "step": 5940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9499298537824972e-05,
+      "loss": 0.9585,
+      "step": 5960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.949596105232077e-05,
+      "loss": 0.9813,
+      "step": 5980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9492612768075094e-05,
+      "loss": 0.9358,
+      "step": 6000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9489253688895594e-05,
+      "loss": 0.8976,
+      "step": 6020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9485883818602194e-05,
+      "loss": 0.9685,
+      "step": 6040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.948250316102709e-05,
+      "loss": 0.9148,
+      "step": 6060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9479111720014737e-05,
+      "loss": 0.8979,
+      "step": 6080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.947570949942186e-05,
+      "loss": 0.914,
+      "step": 6100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9472296503117437e-05,
+      "loss": 0.9587,
+      "step": 6120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9468872734982705e-05,
+      "loss": 0.9355,
+      "step": 6140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9465438198911158e-05,
+      "loss": 0.925,
+      "step": 6160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.946199289880852e-05,
+      "loss": 0.8997,
+      "step": 6180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9458536838592764e-05,
+      "loss": 0.9549,
+      "step": 6200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9455070022194094e-05,
+      "loss": 0.8995,
+      "step": 6220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9451592453554956e-05,
+      "loss": 0.967,
+      "step": 6240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.944810413663001e-05,
+      "loss": 0.945,
+      "step": 6260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9444605075386156e-05,
+      "loss": 0.9658,
+      "step": 6280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.944109527380249e-05,
+      "loss": 0.9514,
+      "step": 6300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9437574735870346e-05,
+      "loss": 0.9264,
+      "step": 6320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9434043465593247e-05,
+      "loss": 0.9749,
+      "step": 6340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9430501466986932e-05,
+      "loss": 0.9507,
+      "step": 6360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9426948744079338e-05,
+      "loss": 0.9251,
+      "step": 6380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.94233853009106e-05,
+      "loss": 0.9631,
+      "step": 6400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9419811141533032e-05,
+      "loss": 0.9639,
+      "step": 6420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9416226270011152e-05,
+      "loss": 0.8744,
+      "step": 6440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9412630690421648e-05,
+      "loss": 0.8919,
+      "step": 6460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.940902440685339e-05,
+      "loss": 0.9569,
+      "step": 6480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9405407423407414e-05,
+      "loss": 0.9714,
+      "step": 6500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9401779744196934e-05,
+      "loss": 0.904,
+      "step": 6520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9398141373347317e-05,
+      "loss": 0.885,
+      "step": 6540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.93944923149961e-05,
+      "loss": 0.9219,
+      "step": 6560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9390832573292964e-05,
+      "loss": 0.9908,
+      "step": 6580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.938716215239974e-05,
+      "loss": 0.8588,
+      "step": 6600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9383481056490402e-05,
+      "loss": 0.8831,
+      "step": 6620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9379789289751073e-05,
+      "loss": 0.9486,
+      "step": 6640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9376086856380007e-05,
+      "loss": 0.9116,
+      "step": 6660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.937237376058758e-05,
+      "loss": 0.941,
+      "step": 6680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.93686500065963e-05,
+      "loss": 0.9913,
+      "step": 6700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9364915598640796e-05,
+      "loss": 0.9316,
+      "step": 6720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.936117054096781e-05,
+      "loss": 0.8926,
+      "step": 6740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9357414837836192e-05,
+      "loss": 0.8829,
+      "step": 6760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.935364849351691e-05,
+      "loss": 0.9302,
+      "step": 6780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.934987151229302e-05,
+      "loss": 0.8974,
+      "step": 6800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.934608389845968e-05,
+      "loss": 0.9036,
+      "step": 6820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9342285656324135e-05,
+      "loss": 0.9103,
+      "step": 6840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9338476790205726e-05,
+      "loss": 0.9448,
+      "step": 6860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.933465730443586e-05,
+      "loss": 0.9723,
+      "step": 6880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9330827203358044e-05,
+      "loss": 0.9088,
+      "step": 6900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.932698649132783e-05,
+      "loss": 0.9263,
+      "step": 6920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.932313517271285e-05,
+      "loss": 0.9501,
+      "step": 6940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9319273251892805e-05,
+      "loss": 0.9005,
+      "step": 6960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.931540073325944e-05,
+      "loss": 0.9459,
+      "step": 6980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9311517621216555e-05,
+      "loss": 0.9368,
+      "step": 7000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.930762392018e-05,
+      "loss": 0.9028,
+      "step": 7020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9303719634577668e-05,
+      "loss": 0.9928,
+      "step": 7040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.929980476884948e-05,
+      "loss": 0.9081,
+      "step": 7060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.92958793274474e-05,
+      "loss": 0.9539,
+      "step": 7080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9291943314835406e-05,
+      "loss": 0.9098,
+      "step": 7100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.928799673548951e-05,
+      "loss": 0.9151,
+      "step": 7120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9284039593897734e-05,
+      "loss": 0.9074,
+      "step": 7140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.928007189456011e-05,
+      "loss": 0.8805,
+      "step": 7160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9276093641988684e-05,
+      "loss": 0.9018,
+      "step": 7180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9272104840707487e-05,
+      "loss": 0.9576,
+      "step": 7200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9268105495252568e-05,
+      "loss": 0.9561,
+      "step": 7220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9264095610171944e-05,
+      "loss": 0.8706,
+      "step": 7240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.926007519002564e-05,
+      "loss": 0.9495,
+      "step": 7260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.925604423938564e-05,
+      "loss": 0.9262,
+      "step": 7280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9252002762835917e-05,
+      "loss": 0.9658,
+      "step": 7300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.924795076497241e-05,
+      "loss": 0.9591,
+      "step": 7320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.924388825040303e-05,
+      "loss": 0.9324,
+      "step": 7340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9239815223747632e-05,
+      "loss": 0.8588,
+      "step": 7360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9235731689638035e-05,
+      "loss": 0.8995,
+      "step": 7380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9231637652718004e-05,
+      "loss": 0.8604,
+      "step": 7400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9227533117643256e-05,
+      "loss": 0.918,
+      "step": 7420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.922341808908144e-05,
+      "loss": 0.9008,
+      "step": 7440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.921929257171213e-05,
+      "loss": 0.9243,
+      "step": 7460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9215156570226844e-05,
+      "loss": 0.9138,
+      "step": 7480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.921101008932901e-05,
+      "loss": 0.9794,
+      "step": 7500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.920685313373398e-05,
+      "loss": 0.8578,
+      "step": 7520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.920268570816901e-05,
+      "loss": 0.9052,
+      "step": 7540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9198507817373274e-05,
+      "loss": 0.9486,
+      "step": 7560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.919431946609784e-05,
+      "loss": 0.9456,
+      "step": 7580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9190120659105663e-05,
+      "loss": 0.9079,
+      "step": 7600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9185911401171615e-05,
+      "loss": 0.9492,
+      "step": 7620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.918169169708242e-05,
+      "loss": 0.9227,
+      "step": 7640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.91774615516367e-05,
+      "loss": 0.9094,
+      "step": 7660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9173220969644947e-05,
+      "loss": 0.8853,
+      "step": 7680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.916896995592953e-05,
+      "loss": 0.9563,
+      "step": 7700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.916470851532466e-05,
+      "loss": 0.9817,
+      "step": 7720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9160436652676425e-05,
+      "loss": 1.0074,
+      "step": 7740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9156154372842757e-05,
+      "loss": 0.9542,
+      "step": 7760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9151861680693432e-05,
+      "loss": 0.8859,
+      "step": 7780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.914755858111008e-05,
+      "loss": 0.9705,
+      "step": 7800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.914324507898614e-05,
+      "loss": 0.8557,
+      "step": 7820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9138921179226908e-05,
+      "loss": 0.8834,
+      "step": 7840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9134586886749492e-05,
+      "loss": 0.9064,
+      "step": 7860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9130242206482817e-05,
+      "loss": 0.8905,
+      "step": 7880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.912588714336762e-05,
+      "loss": 0.88,
+      "step": 7900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.912152170235646e-05,
+      "loss": 0.8668,
+      "step": 7920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9117145888413678e-05,
+      "loss": 0.9504,
+      "step": 7940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.911275970651542e-05,
+      "loss": 0.9257,
+      "step": 7960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.910836316164962e-05,
+      "loss": 0.9762,
+      "step": 7980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9103956258816008e-05,
+      "loss": 0.9331,
+      "step": 8000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.909953900302607e-05,
+      "loss": 0.8791,
+      "step": 8020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.909511139930309e-05,
+      "loss": 0.9448,
+      "step": 8040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9090673452682108e-05,
+      "loss": 0.8983,
+      "step": 8060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9086225168209917e-05,
+      "loss": 0.926,
+      "step": 8080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9081766550945084e-05,
+      "loss": 0.9388,
+      "step": 8100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.907729760595791e-05,
+      "loss": 0.9377,
+      "step": 8120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9072818338330455e-05,
+      "loss": 0.9594,
+      "step": 8140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9068328753156514e-05,
+      "loss": 0.8816,
+      "step": 8160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9063828855541605e-05,
+      "loss": 0.9118,
+      "step": 8180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9059318650602986e-05,
+      "loss": 0.9076,
+      "step": 8200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.905479814346962e-05,
+      "loss": 0.8466,
+      "step": 8220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9050267339282214e-05,
+      "loss": 0.9428,
+      "step": 8240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9045726243193155e-05,
+      "loss": 0.9305,
+      "step": 8260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.904117486036655e-05,
+      "loss": 0.8242,
+      "step": 8280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9036613195978198e-05,
+      "loss": 0.902,
+      "step": 8300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9032041255215596e-05,
+      "loss": 0.8798,
+      "step": 8320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9027459043277923e-05,
+      "loss": 0.9021,
+      "step": 8340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.902286656537603e-05,
+      "loss": 0.8776,
+      "step": 8360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9018263826732467e-05,
+      "loss": 0.9082,
+      "step": 8380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9013650832581424e-05,
+      "loss": 0.8431,
+      "step": 8400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9009027588168773e-05,
+      "loss": 0.8934,
+      "step": 8420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.900439409875203e-05,
+      "loss": 0.9072,
+      "step": 8440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.899975036960037e-05,
+      "loss": 0.9326,
+      "step": 8460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8995096405994614e-05,
+      "loss": 0.9009,
+      "step": 8480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.899043221322721e-05,
+      "loss": 0.972,
+      "step": 8500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.898575779660225e-05,
+      "loss": 0.9269,
+      "step": 8520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8981073161435447e-05,
+      "loss": 0.8519,
+      "step": 8540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.897637831305414e-05,
+      "loss": 0.8876,
+      "step": 8560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8971673256797275e-05,
+      "loss": 0.8401,
+      "step": 8580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8966957998015403e-05,
+      "loss": 0.9571,
+      "step": 8600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.896223254207069e-05,
+      "loss": 0.9073,
+      "step": 8620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.89574968943369e-05,
+      "loss": 0.8836,
+      "step": 8640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.895275106019937e-05,
+      "loss": 0.9202,
+      "step": 8660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8947995045055023e-05,
+      "loss": 0.8841,
+      "step": 8680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.894322885431238e-05,
+      "loss": 0.8863,
+      "step": 8700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8938452493391515e-05,
+      "loss": 1.0004,
+      "step": 8720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8933665967724072e-05,
+      "loss": 0.8558,
+      "step": 8740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.892886928275325e-05,
+      "loss": 0.9001,
+      "step": 8760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.892406244393382e-05,
+      "loss": 0.9562,
+      "step": 8780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.891924545673207e-05,
+      "loss": 0.9127,
+      "step": 8800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.891441832662586e-05,
+      "loss": 0.929,
+      "step": 8820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8909581059104557e-05,
+      "loss": 0.8909,
+      "step": 8840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8904733659669072e-05,
+      "loss": 0.9258,
+      "step": 8860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8899876133831835e-05,
+      "loss": 0.9803,
+      "step": 8880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8895008487116793e-05,
+      "loss": 0.8716,
+      "step": 8900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8890130725059396e-05,
+      "loss": 0.924,
+      "step": 8920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8885242853206606e-05,
+      "loss": 0.8649,
+      "step": 8940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.888034487711687e-05,
+      "loss": 0.9427,
+      "step": 8960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8875436802360142e-05,
+      "loss": 0.9313,
+      "step": 8980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8870518634517843e-05,
+      "loss": 0.9407,
+      "step": 9000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8865590379182885e-05,
+      "loss": 0.9581,
+      "step": 9020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8860652041959643e-05,
+      "loss": 0.8846,
+      "step": 9040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8855703628463957e-05,
+      "loss": 0.8592,
+      "step": 9060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8850745144323137e-05,
+      "loss": 0.9586,
+      "step": 9080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.884577659517593e-05,
+      "loss": 0.9452,
+      "step": 9100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.884079798667254e-05,
+      "loss": 0.8987,
+      "step": 9120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.88358093244746e-05,
+      "loss": 0.925,
+      "step": 9140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.883081061425519e-05,
+      "loss": 0.8806,
+      "step": 9160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.882580186169881e-05,
+      "loss": 0.8575,
+      "step": 9180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.882078307250137e-05,
+      "loss": 0.8711,
+      "step": 9200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8815754252370205e-05,
+      "loss": 0.9003,
+      "step": 9220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.881071540702406e-05,
+      "loss": 0.9115,
+      "step": 9240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8805666542193073e-05,
+      "loss": 0.9562,
+      "step": 9260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8800607663618774e-05,
+      "loss": 0.8856,
+      "step": 9280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.879553877705409e-05,
+      "loss": 0.8638,
+      "step": 9300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8790459888263323e-05,
+      "loss": 0.882,
+      "step": 9320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8785371003022145e-05,
+      "loss": 0.8989,
+      "step": 9340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8780272127117606e-05,
+      "loss": 0.9034,
+      "step": 9360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8775163266348116e-05,
+      "loss": 0.8835,
+      "step": 9380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.877004442652343e-05,
+      "loss": 0.816,
+      "step": 9400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8764915613464657e-05,
+      "loss": 0.8363,
+      "step": 9420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8759776833004256e-05,
+      "loss": 0.869,
+      "step": 9440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8754628090986e-05,
+      "loss": 0.8829,
+      "step": 9460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8749469393265016e-05,
+      "loss": 0.903,
+      "step": 9480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.874430074570773e-05,
+      "loss": 0.9455,
+      "step": 9500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.87391221541919e-05,
+      "loss": 0.9328,
+      "step": 9520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.873393362460658e-05,
+      "loss": 0.9131,
+      "step": 9540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8728735162852125e-05,
+      "loss": 0.9119,
+      "step": 9560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.87235267748402e-05,
+      "loss": 0.8505,
+      "step": 9580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8718308466493745e-05,
+      "loss": 0.8926,
+      "step": 9600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.871308024374698e-05,
+      "loss": 0.8904,
+      "step": 9620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8707842112545406e-05,
+      "loss": 0.8506,
+      "step": 9640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.870259407884579e-05,
+      "loss": 0.903,
+      "step": 9660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8697336148616162e-05,
+      "loss": 0.9225,
+      "step": 9680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.86920683278358e-05,
+      "loss": 0.8845,
+      "step": 9700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.868679062249524e-05,
+      "loss": 0.8764,
+      "step": 9720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8681503038596248e-05,
+      "loss": 0.9878,
+      "step": 9740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8676205582151822e-05,
+      "loss": 0.9531,
+      "step": 9760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8670898259186202e-05,
+      "loss": 0.8972,
+      "step": 9780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8665581075734836e-05,
+      "loss": 0.8453,
+      "step": 9800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.9649,
+      "step": 9820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.865491715157273e-05,
+      "loss": 0.886,
+      "step": 9840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.864957042298893e-05,
+      "loss": 0.8603,
+      "step": 9860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.864421385817325e-05,
+      "loss": 0.9179,
+      "step": 9880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8638847463217145e-05,
+      "loss": 0.8456,
+      "step": 9900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8633471244223232e-05,
+      "loss": 0.8553,
+      "step": 9920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8628085207305325e-05,
+      "loss": 0.8574,
+      "step": 9940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8622689358588373e-05,
+      "loss": 0.8474,
+      "step": 9960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8617283704208507e-05,
+      "loss": 0.905,
+      "step": 9980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8611868250312994e-05,
+      "loss": 0.9268,
+      "step": 10000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.860644300306026e-05,
+      "loss": 0.8962,
+      "step": 10020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8601007968619847e-05,
+      "loss": 0.9645,
+      "step": 10040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8595563153172448e-05,
+      "loss": 0.9035,
+      "step": 10060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8590108562909862e-05,
+      "loss": 0.9684,
+      "step": 10080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.858464420403502e-05,
+      "loss": 0.863,
+      "step": 10100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.857917008276195e-05,
+      "loss": 0.9421,
+      "step": 10120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.857368620531578e-05,
+      "loss": 0.8535,
+      "step": 10140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8568192577932746e-05,
+      "loss": 0.8553,
+      "step": 10160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8562689206860155e-05,
+      "loss": 0.8977,
+      "step": 10180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8557176098356406e-05,
+      "loss": 0.9678,
+      "step": 10200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8551653258690972e-05,
+      "loss": 0.8443,
+      "step": 10220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.854612069414438e-05,
+      "loss": 0.9013,
+      "step": 10240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8540578411008233e-05,
+      "loss": 0.8929,
+      "step": 10260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8535026415585168e-05,
+      "loss": 0.9134,
+      "step": 10280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8529464714188888e-05,
+      "loss": 0.898,
+      "step": 10300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.852389331314411e-05,
+      "loss": 0.8557,
+      "step": 10320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.85183122187866e-05,
+      "loss": 0.9308,
+      "step": 10340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8512721437463136e-05,
+      "loss": 0.8048,
+      "step": 10360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8507120975531517e-05,
+      "loss": 0.8329,
+      "step": 10380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.850151083936055e-05,
+      "loss": 0.9244,
+      "step": 10400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.849589103533004e-05,
+      "loss": 0.8769,
+      "step": 10420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.84902615698308e-05,
+      "loss": 0.8796,
+      "step": 10440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8484622449264608e-05,
+      "loss": 0.8567,
+      "step": 10460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8478973680044237e-05,
+      "loss": 0.8471,
+      "step": 10480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8473315268593432e-05,
+      "loss": 0.8955,
+      "step": 10500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.846764722134689e-05,
+      "loss": 0.8996,
+      "step": 10520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8461969544750286e-05,
+      "loss": 0.8435,
+      "step": 10540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.845628224526023e-05,
+      "loss": 0.862,
+      "step": 10560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8450585329344277e-05,
+      "loss": 0.8905,
+      "step": 10580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8444878803480925e-05,
+      "loss": 0.9191,
+      "step": 10600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8439162674159595e-05,
+      "loss": 0.9315,
+      "step": 10620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8433436947880626e-05,
+      "loss": 0.8691,
+      "step": 10640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.842770163115528e-05,
+      "loss": 0.9158,
+      "step": 10660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.842195673050572e-05,
+      "loss": 0.9328,
+      "step": 10680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8416202252465e-05,
+      "loss": 0.8776,
+      "step": 10700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.841043820357708e-05,
+      "loss": 0.8524,
+      "step": 10720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8404664590396796e-05,
+      "loss": 0.8113,
+      "step": 10740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8398881419489854e-05,
+      "loss": 0.8449,
+      "step": 10760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8393088697432847e-05,
+      "loss": 0.8801,
+      "step": 10780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.838728643081321e-05,
+      "loss": 0.8911,
+      "step": 10800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8381474626229245e-05,
+      "loss": 0.8365,
+      "step": 10820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8375653290290093e-05,
+      "loss": 0.8991,
+      "step": 10840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.836982242961574e-05,
+      "loss": 0.8696,
+      "step": 10860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8363982050836996e-05,
+      "loss": 0.8692,
+      "step": 10880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.835813216059551e-05,
+      "loss": 0.892,
+      "step": 10900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8352272765543723e-05,
+      "loss": 0.8675,
+      "step": 10920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8346403872344907e-05,
+      "loss": 0.8757,
+      "step": 10940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8340525487673128e-05,
+      "loss": 0.942,
+      "step": 10960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8334637618213238e-05,
+      "loss": 0.9497,
+      "step": 10980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8328740270660887e-05,
+      "loss": 0.8416,
+      "step": 11000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.83228334517225e-05,
+      "loss": 0.8326,
+      "step": 11020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.831691716811526e-05,
+      "loss": 0.8522,
+      "step": 11040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8310991426567137e-05,
+      "loss": 0.9059,
+      "step": 11060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8305056233816833e-05,
+      "loss": 0.9037,
+      "step": 11080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8299111596613815e-05,
+      "loss": 0.8691,
+      "step": 11100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8293157521718284e-05,
+      "loss": 0.8107,
+      "step": 11120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8287194015901167e-05,
+      "loss": 0.9504,
+      "step": 11140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8281221085944128e-05,
+      "loss": 0.8615,
+      "step": 11160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.827523873863954e-05,
+      "loss": 0.8942,
+      "step": 11180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8269246980790482e-05,
+      "loss": 0.927,
+      "step": 11200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8263245819210745e-05,
+      "loss": 0.9224,
+      "step": 11220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.825723526072481e-05,
+      "loss": 0.9215,
+      "step": 11240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8251215312167833e-05,
+      "loss": 0.8342,
+      "step": 11260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8245185980385673e-05,
+      "loss": 0.8617,
+      "step": 11280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.823914727223483e-05,
+      "loss": 0.8649,
+      "step": 11300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.823309919458249e-05,
+      "loss": 0.9609,
+      "step": 11320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8227041754306484e-05,
+      "loss": 0.9503,
+      "step": 11340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.822097495829529e-05,
+      "loss": 0.8748,
+      "step": 11360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.821489881344803e-05,
+      "loss": 0.9055,
+      "step": 11380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8208813326674444e-05,
+      "loss": 0.8596,
+      "step": 11400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8202718504894917e-05,
+      "loss": 0.9291,
+      "step": 11420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8196614355040436e-05,
+      "loss": 0.9413,
+      "step": 11440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.819050088405259e-05,
+      "loss": 0.9221,
+      "step": 11460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.818437809888359e-05,
+      "loss": 0.9094,
+      "step": 11480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.817824600649621e-05,
+      "loss": 0.821,
+      "step": 11500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8172104613863838e-05,
+      "loss": 0.9144,
+      "step": 11520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.816595392797041e-05,
+      "loss": 0.8497,
+      "step": 11540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8159793955810453e-05,
+      "loss": 0.9188,
+      "step": 11560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.815362470438904e-05,
+      "loss": 0.8214,
+      "step": 11580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8147446180721802e-05,
+      "loss": 0.9231,
+      "step": 11600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8141258391834914e-05,
+      "loss": 0.8962,
+      "step": 11620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8135061344765088e-05,
+      "loss": 0.8641,
+      "step": 11640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.812885504655956e-05,
+      "loss": 0.8497,
+      "step": 11660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8122639504276096e-05,
+      "loss": 0.8791,
+      "step": 11680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.811641472498296e-05,
+      "loss": 0.8411,
+      "step": 11700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8110180715758934e-05,
+      "loss": 0.9839,
+      "step": 11720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8103937483693288e-05,
+      "loss": 0.9035,
+      "step": 11740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.809768503588578e-05,
+      "loss": 0.8555,
+      "step": 11760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8091423379446658e-05,
+      "loss": 0.8599,
+      "step": 11780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8085152521496627e-05,
+      "loss": 0.8166,
+      "step": 11800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8078872469166864e-05,
+      "loss": 0.8676,
+      "step": 11820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8072583229599004e-05,
+      "loss": 0.8732,
+      "step": 11840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8066284809945128e-05,
+      "loss": 0.9253,
+      "step": 11860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8059977217367753e-05,
+      "loss": 0.918,
+      "step": 11880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8053660459039833e-05,
+      "loss": 0.8492,
+      "step": 11900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8047334542144738e-05,
+      "loss": 0.8387,
+      "step": 11920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8040999473876258e-05,
+      "loss": 0.8212,
+      "step": 11940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8034655261438588e-05,
+      "loss": 0.8189,
+      "step": 11960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8028301912046325e-05,
+      "loss": 0.8555,
+      "step": 11980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8021939432924456e-05,
+      "loss": 0.8558,
+      "step": 12000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8015567831308344e-05,
+      "loss": 0.935,
+      "step": 12020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.800918711444373e-05,
+      "loss": 0.8454,
+      "step": 12040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8002797289586726e-05,
+      "loss": 0.8995,
+      "step": 12060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7996398364003797e-05,
+      "loss": 0.8524,
+      "step": 12080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7989990344971747e-05,
+      "loss": 0.8203,
+      "step": 12100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7983573239777746e-05,
+      "loss": 0.8227,
+      "step": 12120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.797714705571927e-05,
+      "loss": 0.8572,
+      "step": 12140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7970711800104138e-05,
+      "loss": 0.8715,
+      "step": 12160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7964267480250476e-05,
+      "loss": 0.8814,
+      "step": 12180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7957814103486716e-05,
+      "loss": 0.8674,
+      "step": 12200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7951351677151604e-05,
+      "loss": 0.8933,
+      "step": 12220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7944880208594156e-05,
+      "loss": 0.8717,
+      "step": 12240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7938399705173687e-05,
+      "loss": 0.8722,
+      "step": 12260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7931910174259777e-05,
+      "loss": 0.8855,
+      "step": 12280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.792541162323228e-05,
+      "loss": 0.8634,
+      "step": 12300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7918904059481303e-05,
+      "loss": 0.9169,
+      "step": 12320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.79123874904072e-05,
+      "loss": 0.9152,
+      "step": 12340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7905861923420568e-05,
+      "loss": 0.8336,
+      "step": 12360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7899327365942242e-05,
+      "loss": 0.8865,
+      "step": 12380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7892783825403272e-05,
+      "loss": 0.9224,
+      "step": 12400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.788623130924493e-05,
+      "loss": 0.899,
+      "step": 12420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7879669824918685e-05,
+      "loss": 0.8966,
+      "step": 12440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7873099379886217e-05,
+      "loss": 0.896,
+      "step": 12460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7866519981619396e-05,
+      "loss": 0.9448,
+      "step": 12480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7859931637600256e-05,
+      "loss": 0.8598,
+      "step": 12500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7853334355321025e-05,
+      "loss": 0.8833,
+      "step": 12520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.784672814228409e-05,
+      "loss": 0.8542,
+      "step": 12540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.784011300600198e-05,
+      "loss": 0.8485,
+      "step": 12560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.783348895399739e-05,
+      "loss": 0.9221,
+      "step": 12580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7826855993803148e-05,
+      "loss": 0.8561,
+      "step": 12600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7820214132962205e-05,
+      "loss": 0.8166,
+      "step": 12620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7813563379027644e-05,
+      "loss": 0.8632,
+      "step": 12640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7806903739562652e-05,
+      "loss": 0.8326,
+      "step": 12660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7800235222140525e-05,
+      "loss": 0.9188,
+      "step": 12680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7793557834344662e-05,
+      "loss": 0.8959,
+      "step": 12700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7786871583768536e-05,
+      "loss": 0.796,
+      "step": 12720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7780176478015706e-05,
+      "loss": 0.7971,
+      "step": 12740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7773472524699803e-05,
+      "loss": 0.9084,
+      "step": 12760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7766759731444512e-05,
+      "loss": 0.8869,
+      "step": 12780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.776003810588358e-05,
+      "loss": 0.8769,
+      "step": 12800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7753307655660792e-05,
+      "loss": 0.8849,
+      "step": 12820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7746568388429966e-05,
+      "loss": 0.8451,
+      "step": 12840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7739820311854957e-05,
+      "loss": 0.9419,
+      "step": 12860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7733063433609623e-05,
+      "loss": 0.9293,
+      "step": 12880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7726297761377847e-05,
+      "loss": 0.8162,
+      "step": 12900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7719523302853498e-05,
+      "loss": 0.8825,
+      "step": 12920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.771274006574045e-05,
+      "loss": 0.8961,
+      "step": 12940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7705948057752547e-05,
+      "loss": 0.9109,
+      "step": 12960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.769914728661362e-05,
+      "loss": 0.8448,
+      "step": 12980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7692337760057454e-05,
+      "loss": 0.9286,
+      "step": 13000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.76855194858278e-05,
+      "loss": 0.9075,
+      "step": 13020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.767869247167835e-05,
+      "loss": 0.8712,
+      "step": 13040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7671856725372737e-05,
+      "loss": 0.8144,
+      "step": 13060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7665012254684522e-05,
+      "loss": 0.8475,
+      "step": 13080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.76581590673972e-05,
+      "loss": 0.9759,
+      "step": 13100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7651297171304157e-05,
+      "loss": 0.821,
+      "step": 13120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7644426574208698e-05,
+      "loss": 0.8912,
+      "step": 13140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.763754728392402e-05,
+      "loss": 0.9006,
+      "step": 13160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7630659308273204e-05,
+      "loss": 0.87,
+      "step": 13180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7623762655089208e-05,
+      "loss": 0.9281,
+      "step": 13200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.761685733221486e-05,
+      "loss": 0.8586,
+      "step": 13220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7609943347502846e-05,
+      "loss": 0.8539,
+      "step": 13240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7603020708815695e-05,
+      "loss": 0.8956,
+      "step": 13260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7596089424025794e-05,
+      "loss": 0.9335,
+      "step": 13280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7589149501015345e-05,
+      "loss": 0.8925,
+      "step": 13300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.758220094767638e-05,
+      "loss": 0.8832,
+      "step": 13320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.757524377191075e-05,
+      "loss": 0.8692,
+      "step": 13340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7568277981630107e-05,
+      "loss": 0.9298,
+      "step": 13360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7561303584755903e-05,
+      "loss": 0.9228,
+      "step": 13380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7554320589219368e-05,
+      "loss": 0.8039,
+      "step": 13400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.754732900296151e-05,
+      "loss": 0.9091,
+      "step": 13420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.754032883393313e-05,
+      "loss": 0.9266,
+      "step": 13440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7533320090094763e-05,
+      "loss": 0.8973,
+      "step": 13460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7526302779416703e-05,
+      "loss": 0.8943,
+      "step": 13480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7519276909878988e-05,
+      "loss": 0.9595,
+      "step": 13500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.751224248947139e-05,
+      "loss": 0.9303,
+      "step": 13520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7505199526193398e-05,
+      "loss": 0.7938,
+      "step": 13540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.749814802805423e-05,
+      "loss": 0.8912,
+      "step": 13560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7491088003072798e-05,
+      "loss": 0.908,
+      "step": 13580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.748401945927771e-05,
+      "loss": 0.9292,
+      "step": 13600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7476942404707267e-05,
+      "loss": 0.904,
+      "step": 13620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.746985684740945e-05,
+      "loss": 0.8896,
+      "step": 13640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.74627627954419e-05,
+      "loss": 0.8471,
+      "step": 13660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.745566025687193e-05,
+      "loss": 0.8564,
+      "step": 13680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.74485492397765e-05,
+      "loss": 0.8961,
+      "step": 13700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.74414297522422e-05,
+      "loss": 0.8472,
+      "step": 13720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7434301802365264e-05,
+      "loss": 0.8185,
+      "step": 13740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7427165398251557e-05,
+      "loss": 0.8764,
+      "step": 13760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7420020548016538e-05,
+      "loss": 0.912,
+      "step": 13780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7412867259785286e-05,
+      "loss": 0.8544,
+      "step": 13800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7405705541692476e-05,
+      "loss": 0.9203,
+      "step": 13820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7398535401882353e-05,
+      "loss": 0.83,
+      "step": 13840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.739135684850876e-05,
+      "loss": 0.9542,
+      "step": 13860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7384169889735094e-05,
+      "loss": 0.8845,
+      "step": 13880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7376974533734317e-05,
+      "loss": 0.9061,
+      "step": 13900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7369770788688937e-05,
+      "loss": 0.9571,
+      "step": 13920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7362558662791003e-05,
+      "loss": 0.8479,
+      "step": 13940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7355338164242098e-05,
+      "loss": 0.8484,
+      "step": 13960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.734810930125332e-05,
+      "loss": 0.862,
+      "step": 13980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.734087208204528e-05,
+      "loss": 0.8531,
+      "step": 14000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.73336265148481e-05,
+      "loss": 0.8025,
+      "step": 14020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7326372607901388e-05,
+      "loss": 0.8492,
+      "step": 14040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7319110369454232e-05,
+      "loss": 0.8061,
+      "step": 14060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7311839807765204e-05,
+      "loss": 0.8537,
+      "step": 14080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7304560931102334e-05,
+      "loss": 0.8739,
+      "step": 14100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.729727374774311e-05,
+      "loss": 0.8787,
+      "step": 14120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7289978265974466e-05,
+      "loss": 0.8357,
+      "step": 14140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.728267449409278e-05,
+      "loss": 0.8385,
+      "step": 14160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7275362440403844e-05,
+      "loss": 0.9139,
+      "step": 14180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7268042113222877e-05,
+      "loss": 0.9294,
+      "step": 14200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.726071352087451e-05,
+      "loss": 0.9138,
+      "step": 14220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.725337667169276e-05,
+      "loss": 0.8542,
+      "step": 14240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7246031574021047e-05,
+      "loss": 0.8184,
+      "step": 14260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.723867823621216e-05,
+      "loss": 0.8665,
+      "step": 14280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7231316666628268e-05,
+      "loss": 0.8361,
+      "step": 14300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7223946873640897e-05,
+      "loss": 0.8156,
+      "step": 14320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7216568865630924e-05,
+      "loss": 0.8673,
+      "step": 14340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.720918265098857e-05,
+      "loss": 0.8501,
+      "step": 14360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.720178823811338e-05,
+      "loss": 0.8468,
+      "step": 14380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7194385635414242e-05,
+      "loss": 0.8028,
+      "step": 14400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.718697485130934e-05,
+      "loss": 0.837,
+      "step": 14420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.717955589422616e-05,
+      "loss": 0.8196,
+      "step": 14440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7172128772601497e-05,
+      "loss": 0.932,
+      "step": 14460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7164693494881418e-05,
+      "loss": 0.85,
+      "step": 14480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.715725006952127e-05,
+      "loss": 0.8667,
+      "step": 14500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7149798504985665e-05,
+      "loss": 0.8423,
+      "step": 14520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7142338809748475e-05,
+      "loss": 0.9186,
+      "step": 14540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7134870992292808e-05,
+      "loss": 0.8717,
+      "step": 14560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.712739506111102e-05,
+      "loss": 0.8056,
+      "step": 14580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7119911024704684e-05,
+      "loss": 0.9074,
+      "step": 14600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7112418891584598e-05,
+      "loss": 0.867,
+      "step": 14620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7104918670270763e-05,
+      "loss": 0.8923,
+      "step": 14640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7097410369292376e-05,
+      "loss": 0.915,
+      "step": 14660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.708989399718783e-05,
+      "loss": 0.9823,
+      "step": 14680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.708236956250469e-05,
+      "loss": 0.8399,
+      "step": 14700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.707483707379969e-05,
+      "loss": 0.8465,
+      "step": 14720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.706729653963872e-05,
+      "loss": 0.922,
+      "step": 14740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7059747968596836e-05,
+      "loss": 0.8394,
+      "step": 14760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.705219136925821e-05,
+      "loss": 0.7721,
+      "step": 14780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.704462675021616e-05,
+      "loss": 0.7611,
+      "step": 14800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7037054120073115e-05,
+      "loss": 0.8898,
+      "step": 14820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7029473487440617e-05,
+      "loss": 0.8221,
+      "step": 14840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7021884860939314e-05,
+      "loss": 0.9067,
+      "step": 14860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7014288249198937e-05,
+      "loss": 0.8242,
+      "step": 14880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.70066836608583e-05,
+      "loss": 0.8687,
+      "step": 14900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.699907110456529e-05,
+      "loss": 0.917,
+      "step": 14920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6991450588976853e-05,
+      "loss": 0.8522,
+      "step": 14940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.698382212275899e-05,
+      "loss": 0.8023,
+      "step": 14960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.697618571458673e-05,
+      "loss": 0.8668,
+      "step": 14980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6968541373144157e-05,
+      "loss": 0.8549,
+      "step": 15000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.696088910712435e-05,
+      "loss": 0.851,
+      "step": 15020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.695322892522943e-05,
+      "loss": 0.9246,
+      "step": 15040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6945560836170485e-05,
+      "loss": 0.9117,
+      "step": 15060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6937884848667625e-05,
+      "loss": 0.8651,
+      "step": 15080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6930200971449923e-05,
+      "loss": 0.8053,
+      "step": 15100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.692250921325544e-05,
+      "loss": 0.8642,
+      "step": 15120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.691480958283118e-05,
+      "loss": 0.7826,
+      "step": 15140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6907102088933118e-05,
+      "loss": 0.8057,
+      "step": 15160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6899386740326164e-05,
+      "loss": 0.8133,
+      "step": 15180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6891663545784146e-05,
+      "loss": 0.8651,
+      "step": 15200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6883932514089842e-05,
+      "loss": 0.8253,
+      "step": 15220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6876193654034922e-05,
+      "loss": 0.9001,
+      "step": 15240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6868446974419962e-05,
+      "loss": 0.8583,
+      "step": 15260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.686069248405444e-05,
+      "loss": 0.8354,
+      "step": 15280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.68529301917567e-05,
+      "loss": 0.8794,
+      "step": 15300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6845160106353975e-05,
+      "loss": 0.8645,
+      "step": 15320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6837382236682347e-05,
+      "loss": 0.8513,
+      "step": 15340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.682959659158676e-05,
+      "loss": 0.8559,
+      "step": 15360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6821803179920996e-05,
+      "loss": 0.7707,
+      "step": 15380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.681400201054766e-05,
+      "loss": 0.7838,
+      "step": 15400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.68061930923382e-05,
+      "loss": 0.8378,
+      "step": 15420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6798376434172862e-05,
+      "loss": 0.8646,
+      "step": 15440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6790552044940692e-05,
+      "loss": 0.8471,
+      "step": 15460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.678271993353953e-05,
+      "loss": 0.8338,
+      "step": 15480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6774880108876005e-05,
+      "loss": 0.8486,
+      "step": 15500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6767032579865503e-05,
+      "loss": 0.8262,
+      "step": 15520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6759177355432186e-05,
+      "loss": 0.8425,
+      "step": 15540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6751314444508967e-05,
+      "loss": 0.8316,
+      "step": 15560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.674344385603748e-05,
+      "loss": 0.8687,
+      "step": 15580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6735565598968114e-05,
+      "loss": 0.8748,
+      "step": 15600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.672767968225997e-05,
+      "loss": 0.8936,
+      "step": 15620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6719786114880846e-05,
+      "loss": 0.8964,
+      "step": 15640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6711884905807264e-05,
+      "loss": 0.8166,
+      "step": 15660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.670397606402442e-05,
+      "loss": 0.8327,
+      "step": 15680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.66960595985262e-05,
+      "loss": 0.8051,
+      "step": 15700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6688135518315144e-05,
+      "loss": 0.8647,
+      "step": 15720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.668020383240247e-05,
+      "loss": 0.8963,
+      "step": 15740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6672264549808033e-05,
+      "loss": 0.8542,
+      "step": 15760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6664317679560333e-05,
+      "loss": 0.8557,
+      "step": 15780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.665636323069649e-05,
+      "loss": 0.8623,
+      "step": 15800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6648401212262253e-05,
+      "loss": 0.8469,
+      "step": 15820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6640431633311975e-05,
+      "loss": 0.8281,
+      "step": 15840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6632454502908605e-05,
+      "loss": 0.9213,
+      "step": 15860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.662446983012368e-05,
+      "loss": 0.8172,
+      "step": 15880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6616477624037317e-05,
+      "loss": 0.9342,
+      "step": 15900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6608477893738197e-05,
+      "loss": 0.8882,
+      "step": 15920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6600470648323554e-05,
+      "loss": 0.8612,
+      "step": 15940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6592455896899177e-05,
+      "loss": 0.7878,
+      "step": 15960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.658443364857939e-05,
+      "loss": 0.8418,
+      "step": 15980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6576403912487026e-05,
+      "loss": 0.86,
+      "step": 16000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.656836669775346e-05,
+      "loss": 0.9031,
+      "step": 16020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6560322013518542e-05,
+      "loss": 0.7993,
+      "step": 16040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6552269868930646e-05,
+      "loss": 0.9313,
+      "step": 16060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6544210273146608e-05,
+      "loss": 0.8022,
+      "step": 16080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6536143235331743e-05,
+      "loss": 0.8633,
+      "step": 16100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6528068764659838e-05,
+      "loss": 0.8447,
+      "step": 16120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.651998687031312e-05,
+      "loss": 0.8093,
+      "step": 16140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6511897561482263e-05,
+      "loss": 0.8275,
+      "step": 16160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6503800847366373e-05,
+      "loss": 1.0032,
+      "step": 16180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6495696737172982e-05,
+      "loss": 0.8871,
+      "step": 16200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6487585240118025e-05,
+      "loss": 0.8384,
+      "step": 16220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.647946636542584e-05,
+      "loss": 0.8278,
+      "step": 16240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6471340122329158e-05,
+      "loss": 0.8858,
+      "step": 16260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6463206520069083e-05,
+      "loss": 0.8531,
+      "step": 16280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6455065567895098e-05,
+      "loss": 0.8399,
+      "step": 16300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6446917275065032e-05,
+      "loss": 0.8538,
+      "step": 16320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.643876165084507e-05,
+      "loss": 0.7951,
+      "step": 16340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6430598704509723e-05,
+      "loss": 0.8208,
+      "step": 16360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.642242844534185e-05,
+      "loss": 0.8307,
+      "step": 16380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6414250882632607e-05,
+      "loss": 0.8354,
+      "step": 16400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6406066025681462e-05,
+      "loss": 0.9212,
+      "step": 16420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6397873883796183e-05,
+      "loss": 0.8434,
+      "step": 16440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6389674466292812e-05,
+      "loss": 0.8562,
+      "step": 16460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.638146778249567e-05,
+      "loss": 0.8936,
+      "step": 16480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.637325384173734e-05,
+      "loss": 0.833,
+      "step": 16500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6365032653358665e-05,
+      "loss": 0.7673,
+      "step": 16520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.635680422670872e-05,
+      "loss": 0.8474,
+      "step": 16540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6348568571144816e-05,
+      "loss": 0.9436,
+      "step": 16560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.634032569603248e-05,
+      "loss": 0.8744,
+      "step": 16580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.633207561074546e-05,
+      "loss": 0.8673,
+      "step": 16600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6323818324665693e-05,
+      "loss": 0.827,
+      "step": 16620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.63155538471833e-05,
+      "loss": 0.8584,
+      "step": 16640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6307282187696602e-05,
+      "loss": 0.8387,
+      "step": 16660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.629900335561206e-05,
+      "loss": 0.8471,
+      "step": 16680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6290717360344312e-05,
+      "loss": 0.7533,
+      "step": 16700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6282424211316134e-05,
+      "loss": 0.8486,
+      "step": 16720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6274123917958438e-05,
+      "loss": 0.8284,
+      "step": 16740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.626581648971026e-05,
+      "loss": 0.8575,
+      "step": 16760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.625750193601875e-05,
+      "loss": 0.8387,
+      "step": 16780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.624918026633916e-05,
+      "loss": 0.8599,
+      "step": 16800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6240851490134834e-05,
+      "loss": 0.8463,
+      "step": 16820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6232515616877204e-05,
+      "loss": 0.8325,
+      "step": 16840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6224172656045767e-05,
+      "loss": 0.8539,
+      "step": 16860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6215822617128075e-05,
+      "loss": 0.8314,
+      "step": 16880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.620746550961974e-05,
+      "loss": 0.8184,
+      "step": 16900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6199101343024405e-05,
+      "loss": 0.838,
+      "step": 16920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.619073012685374e-05,
+      "loss": 0.8426,
+      "step": 16940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.618235187062744e-05,
+      "loss": 0.7958,
+      "step": 16960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6173966583873195e-05,
+      "loss": 0.8954,
+      "step": 16980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6165574276126702e-05,
+      "loss": 0.8166,
+      "step": 17000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6157174956931627e-05,
+      "loss": 0.8335,
+      "step": 17020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6148768635839623e-05,
+      "loss": 0.8346,
+      "step": 17040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6140355322410304e-05,
+      "loss": 0.871,
+      "step": 17060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6131935026211224e-05,
+      "loss": 0.8357,
+      "step": 17080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.612350775681789e-05,
+      "loss": 0.802,
+      "step": 17100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.611507352381373e-05,
+      "loss": 0.8244,
+      "step": 17120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6106632336790105e-05,
+      "loss": 0.7501,
+      "step": 17140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.609818420534627e-05,
+      "loss": 0.792,
+      "step": 17160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.608972913908938e-05,
+      "loss": 0.7854,
+      "step": 17180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.608126714763448e-05,
+      "loss": 0.8522,
+      "step": 17200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6072798240604487e-05,
+      "loss": 0.8495,
+      "step": 17220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6064322427630186e-05,
+      "loss": 0.9158,
+      "step": 17240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6055839718350207e-05,
+      "loss": 0.8612,
+      "step": 17260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6047350122411038e-05,
+      "loss": 0.8484,
+      "step": 17280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6038853649466983e-05,
+      "loss": 0.8433,
+      "step": 17300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6030350309180173e-05,
+      "loss": 0.8917,
+      "step": 17320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6021840111220548e-05,
+      "loss": 0.7789,
+      "step": 17340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6013323065265848e-05,
+      "loss": 0.8575,
+      "step": 17360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.60047991810016e-05,
+      "loss": 0.812,
+      "step": 17380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.59962684681211e-05,
+      "loss": 0.828,
+      "step": 17400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5987730936325424e-05,
+      "loss": 0.8474,
+      "step": 17420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.597918659532339e-05,
+      "loss": 0.8686,
+      "step": 17440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5970635454831565e-05,
+      "loss": 0.7824,
+      "step": 17460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5962077524574248e-05,
+      "loss": 0.9101,
+      "step": 17480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.595351281428345e-05,
+      "loss": 0.8496,
+      "step": 17500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5944941333698912e-05,
+      "loss": 0.8163,
+      "step": 17520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5936363092568058e-05,
+      "loss": 0.854,
+      "step": 17540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5927778100646007e-05,
+      "loss": 0.8085,
+      "step": 17560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5919186367695546e-05,
+      "loss": 0.8276,
+      "step": 17580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5910587903487142e-05,
+      "loss": 0.8778,
+      "step": 17600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5901982717798903e-05,
+      "loss": 0.8123,
+      "step": 17620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5893370820416594e-05,
+      "loss": 0.8197,
+      "step": 17640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5884752221133597e-05,
+      "loss": 0.8753,
+      "step": 17660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.587612692975093e-05,
+      "loss": 0.837,
+      "step": 17680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5867494956077215e-05,
+      "loss": 0.7968,
+      "step": 17700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.585885630992867e-05,
+      "loss": 0.8335,
+      "step": 17720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.58502110011291e-05,
+      "loss": 0.8233,
+      "step": 17740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5841559039509898e-05,
+      "loss": 0.8636,
+      "step": 17760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5832900434910014e-05,
+      "loss": 0.7803,
+      "step": 17780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5824235197175954e-05,
+      "loss": 0.8154,
+      "step": 17800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5815563336161756e-05,
+      "loss": 0.8779,
+      "step": 17820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.580688486172902e-05,
+      "loss": 0.8222,
+      "step": 17840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5798199783746828e-05,
+      "loss": 0.8423,
+      "step": 17860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5789508112091804e-05,
+      "loss": 0.7625,
+      "step": 17880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.578080985664805e-05,
+      "loss": 0.8077,
+      "step": 17900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5772105027307168e-05,
+      "loss": 0.8791,
+      "step": 17920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5763393633968223e-05,
+      "loss": 0.8595,
+      "step": 17940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5754675686537756e-05,
+      "loss": 0.8115,
+      "step": 17960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5745951194929756e-05,
+      "loss": 0.8787,
+      "step": 17980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5737220169065656e-05,
+      "loss": 0.7391,
+      "step": 18000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5728482618874314e-05,
+      "loss": 0.7967,
+      "step": 18020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5719738554292017e-05,
+      "loss": 0.7753,
+      "step": 18040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.571098798526245e-05,
+      "loss": 0.8517,
+      "step": 18060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5702230921736705e-05,
+      "loss": 0.7999,
+      "step": 18080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.569346737367325e-05,
+      "loss": 0.8186,
+      "step": 18100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5684697351037935e-05,
+      "loss": 0.8217,
+      "step": 18120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5675920863803967e-05,
+      "loss": 0.8555,
+      "step": 18140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.566713792195191e-05,
+      "loss": 0.8466,
+      "step": 18160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.565834853546966e-05,
+      "loss": 0.8224,
+      "step": 18180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5649552714352448e-05,
+      "loss": 0.8254,
+      "step": 18200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5640750468602822e-05,
+      "loss": 0.8414,
+      "step": 18220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.563194180823064e-05,
+      "loss": 0.7856,
+      "step": 18240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.562312674325304e-05,
+      "loss": 0.9037,
+      "step": 18260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5614305283694462e-05,
+      "loss": 0.8079,
+      "step": 18280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.56054774395866e-05,
+      "loss": 0.8576,
+      "step": 18300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.559664322096843e-05,
+      "loss": 0.8414,
+      "step": 18320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5587802637886156e-05,
+      "loss": 0.829,
+      "step": 18340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.557895570039323e-05,
+      "loss": 0.8201,
+      "step": 18360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5570102418550327e-05,
+      "loss": 0.7774,
+      "step": 18380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5561242802425337e-05,
+      "loss": 0.8544,
+      "step": 18400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5552376862093357e-05,
+      "loss": 0.8639,
+      "step": 18420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.554350460763667e-05,
+      "loss": 0.8273,
+      "step": 18440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5534626049144753e-05,
+      "loss": 0.7759,
+      "step": 18460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.552574119671423e-05,
+      "loss": 0.8248,
+      "step": 18480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.55168500604489e-05,
+      "loss": 0.8449,
+      "step": 18500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5507952650459697e-05,
+      "loss": 0.7627,
+      "step": 18520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.54990489768647e-05,
+      "loss": 0.7437,
+      "step": 18540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5490139049789097e-05,
+      "loss": 0.8273,
+      "step": 18560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5481222879365205e-05,
+      "loss": 0.8037,
+      "step": 18580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5472300475732427e-05,
+      "loss": 0.7735,
+      "step": 18600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.546337184903726e-05,
+      "loss": 0.8053,
+      "step": 18620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5454437009433277e-05,
+      "loss": 0.8705,
+      "step": 18640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5445495967081114e-05,
+      "loss": 0.8378,
+      "step": 18660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.543654873214846e-05,
+      "loss": 0.8337,
+      "step": 18680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.542759531481006e-05,
+      "loss": 0.766,
+      "step": 18700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5418635725247666e-05,
+      "loss": 0.8558,
+      "step": 18720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.540966997365007e-05,
+      "loss": 0.8342,
+      "step": 18740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.540069807021306e-05,
+      "loss": 0.8172,
+      "step": 18760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.539172002513942e-05,
+      "loss": 0.8388,
+      "step": 18780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.538273584863893e-05,
+      "loss": 0.8054,
+      "step": 18800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.537374555092832e-05,
+      "loss": 0.846,
+      "step": 18820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5364749142231302e-05,
+      "loss": 0.8163,
+      "step": 18840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5355746632778533e-05,
+      "loss": 0.8236,
+      "step": 18860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5346738032807603e-05,
+      "loss": 0.7648,
+      "step": 18880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5337723352563028e-05,
+      "loss": 0.8125,
+      "step": 18900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5328702602296238e-05,
+      "loss": 0.8316,
+      "step": 18920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.531967579226557e-05,
+      "loss": 0.8042,
+      "step": 18940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5310642932736253e-05,
+      "loss": 0.8454,
+      "step": 18960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5301604033980387e-05,
+      "loss": 0.7541,
+      "step": 18980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.529255910627695e-05,
+      "loss": 0.8904,
+      "step": 19000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5283508159911776e-05,
+      "loss": 0.7874,
+      "step": 19020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5274451205177526e-05,
+      "loss": 0.7914,
+      "step": 19040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5265388252373712e-05,
+      "loss": 0.8695,
+      "step": 19060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.525631931180667e-05,
+      "loss": 0.8793,
+      "step": 19080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.524724439378953e-05,
+      "loss": 0.8213,
+      "step": 19100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5238163508642219e-05,
+      "loss": 0.8038,
+      "step": 19120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5229076666691471e-05,
+      "loss": 0.799,
+      "step": 19140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5219983878270767e-05,
+      "loss": 0.8665,
+      "step": 19160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5210885153720374e-05,
+      "loss": 0.8425,
+      "step": 19180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.520178050338729e-05,
+      "loss": 0.8568,
+      "step": 19200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.519266993762527e-05,
+      "loss": 0.8743,
+      "step": 19220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5183553466794778e-05,
+      "loss": 0.8136,
+      "step": 19240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5174431101263004e-05,
+      "loss": 0.848,
+      "step": 19260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5165302851403843e-05,
+      "loss": 0.8114,
+      "step": 19280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5156168727597873e-05,
+      "loss": 0.8329,
+      "step": 19300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5147028740232361e-05,
+      "loss": 0.7901,
+      "step": 19320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5137882899701234e-05,
+      "loss": 0.8436,
+      "step": 19340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5128731216405084e-05,
+      "loss": 0.864,
+      "step": 19360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5119573700751133e-05,
+      "loss": 0.8219,
+      "step": 19380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5110410363153256e-05,
+      "loss": 0.8108,
+      "step": 19400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5101241214031928e-05,
+      "loss": 0.8013,
+      "step": 19420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5092066263814245e-05,
+      "loss": 0.8176,
+      "step": 19440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5082885522933901e-05,
+      "loss": 0.8276,
+      "step": 19460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5073699001831168e-05,
+      "loss": 0.7847,
+      "step": 19480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.506450671095289e-05,
+      "loss": 0.798,
+      "step": 19500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5055308660752486e-05,
+      "loss": 0.8289,
+      "step": 19520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5046104861689911e-05,
+      "loss": 0.8235,
+      "step": 19540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5036895324231662e-05,
+      "loss": 0.8415,
+      "step": 19560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5027680058850757e-05,
+      "loss": 0.8438,
+      "step": 19580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.501845907602674e-05,
+      "loss": 0.7979,
+      "step": 19600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5009232386245644e-05,
+      "loss": 0.8477,
+      "step": 19620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.7946,
+      "step": 19640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4990761927788815e-05,
+      "loss": 0.827,
+      "step": 19660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4981518180117558e-05,
+      "loss": 0.8489,
+      "step": 19680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.497226876749815e-05,
+      "loss": 0.8442,
+      "step": 19700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4963013700448969e-05,
+      "loss": 0.7982,
+      "step": 19720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4953752989494815e-05,
+      "loss": 0.7978,
+      "step": 19740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4944486645166892e-05,
+      "loss": 0.8114,
+      "step": 19760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4935214678002834e-05,
+      "loss": 0.8342,
+      "step": 19780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4925937098546652e-05,
+      "loss": 0.7771,
+      "step": 19800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4916653917348751e-05,
+      "loss": 0.9235,
+      "step": 19820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4907365144965896e-05,
+      "loss": 0.7855,
+      "step": 19840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4898070791961215e-05,
+      "loss": 0.8699,
+      "step": 19860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.4888770868904185e-05,
+      "loss": 0.8658,
+      "step": 19880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4879465386370613e-05,
+      "loss": 0.8394,
+      "step": 19900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.487015435494263e-05,
+      "loss": 0.8058,
+      "step": 19920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4860837785208678e-05,
+      "loss": 0.8074,
+      "step": 19940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4851515687763492e-05,
+      "loss": 0.8149,
+      "step": 19960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.48421880732081e-05,
+      "loss": 0.7852,
+      "step": 19980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.48328549521498e-05,
+      "loss": 0.8215,
+      "step": 20000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4823516335202152e-05,
+      "loss": 0.8319,
+      "step": 20020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4814172232984968e-05,
+      "loss": 0.799,
+      "step": 20040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4804822656124302e-05,
+      "loss": 0.8033,
+      "step": 20060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.479546761525242e-05,
+      "loss": 0.8247,
+      "step": 20080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4786107121007813e-05,
+      "loss": 0.7516,
+      "step": 20100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4776741184035171e-05,
+      "loss": 0.8348,
+      "step": 20120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.476736981498537e-05,
+      "loss": 0.8229,
+      "step": 20140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.475799302451547e-05,
+      "loss": 0.8529,
+      "step": 20160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.4748610823288693e-05,
+      "loss": 0.8749,
+      "step": 20180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4739223221974415e-05,
+      "loss": 0.8914,
+      "step": 20200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4729830231248143e-05,
+      "loss": 0.8839,
+      "step": 20220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4720431861791531e-05,
+      "loss": 0.8345,
+      "step": 20240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4711028124292332e-05,
+      "loss": 0.8047,
+      "step": 20260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4701619029444419e-05,
+      "loss": 0.8083,
+      "step": 20280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4692204587947745e-05,
+      "loss": 0.8185,
+      "step": 20300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.468278481050835e-05,
+      "loss": 0.8176,
+      "step": 20320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4673359707838338e-05,
+      "loss": 0.8108,
+      "step": 20340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4663929290655872e-05,
+      "loss": 0.8012,
+      "step": 20360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4654493569685157e-05,
+      "loss": 0.8082,
+      "step": 20380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.464505255565643e-05,
+      "loss": 0.833,
+      "step": 20400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4635606259305948e-05,
+      "loss": 0.819,
+      "step": 20420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4626154691375974e-05,
+      "loss": 0.8406,
+      "step": 20440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.4616697862614766e-05,
+      "loss": 0.8467,
+      "step": 20460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4607235783776563e-05,
+      "loss": 0.8298,
+      "step": 20480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4597768465621575e-05,
+      "loss": 0.7843,
+      "step": 20500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4588295918915978e-05,
+      "loss": 0.8426,
+      "step": 20520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4578818154431878e-05,
+      "loss": 0.7621,
+      "step": 20540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4569335182947325e-05,
+      "loss": 0.8628,
+      "step": 20560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4559847015246294e-05,
+      "loss": 0.7603,
+      "step": 20580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4550353662118659e-05,
+      "loss": 0.7862,
+      "step": 20600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4540855134360198e-05,
+      "loss": 0.7932,
+      "step": 20620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.453135144277257e-05,
+      "loss": 0.7972,
+      "step": 20640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4521842598163309e-05,
+      "loss": 0.8247,
+      "step": 20660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4512328611345808e-05,
+      "loss": 0.8215,
+      "step": 20680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4502809493139305e-05,
+      "loss": 0.7678,
+      "step": 20700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4493285254368878e-05,
+      "loss": 0.8288,
+      "step": 20720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4483755905865423e-05,
+      "loss": 0.7892,
+      "step": 20740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4474221458465651e-05,
+      "loss": 0.875,
+      "step": 20760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4464681923012072e-05,
+      "loss": 0.7519,
+      "step": 20780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4455137310352978e-05,
+      "loss": 0.8257,
+      "step": 20800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4445587631342433e-05,
+      "loss": 0.8261,
+      "step": 20820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4436032896840271e-05,
+      "loss": 0.8359,
+      "step": 20840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4426473117712071e-05,
+      "loss": 0.8074,
+      "step": 20860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4416908304829142e-05,
+      "loss": 0.8617,
+      "step": 20880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4407338469068532e-05,
+      "loss": 0.7763,
+      "step": 20900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4397763621312985e-05,
+      "loss": 0.8333,
+      "step": 20920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4388183772450957e-05,
+      "loss": 0.8397,
+      "step": 20940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.437859893337658e-05,
+      "loss": 0.8307,
+      "step": 20960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4369009114989671e-05,
+      "loss": 0.8456,
+      "step": 20980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4359414328195704e-05,
+      "loss": 0.8202,
+      "step": 21000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4349814583905807e-05,
+      "loss": 0.7784,
+      "step": 21020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.434020989303674e-05,
+      "loss": 0.8052,
+      "step": 21040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4330600266510895e-05,
+      "loss": 0.7646,
+      "step": 21060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4320985715256269e-05,
+      "loss": 0.8493,
+      "step": 21080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4311366250206464e-05,
+      "loss": 0.7595,
+      "step": 21100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4301741882300673e-05,
+      "loss": 0.7798,
+      "step": 21120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4292112622483653e-05,
+      "loss": 0.827,
+      "step": 21140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4282478481705742e-05,
+      "loss": 0.9012,
+      "step": 21160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4272839470922812e-05,
+      "loss": 0.7908,
+      "step": 21180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.426319560109628e-05,
+      "loss": 0.8161,
+      "step": 21200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4253546883193093e-05,
+      "loss": 0.8432,
+      "step": 21220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.42438933281857e-05,
+      "loss": 0.797,
+      "step": 21240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4234234947052063e-05,
+      "loss": 0.8394,
+      "step": 21260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4224571750775625e-05,
+      "loss": 0.8191,
+      "step": 21280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4214903750345298e-05,
+      "loss": 0.7181,
+      "step": 21300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4205230956755479e-05,
+      "loss": 0.8326,
+      "step": 21320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.419555338100599e-05,
+      "loss": 0.8444,
+      "step": 21340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4185871034102117e-05,
+      "loss": 0.7624,
+      "step": 21360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4176183927054545e-05,
+      "loss": 0.8136,
+      "step": 21380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4166492070879389e-05,
+      "loss": 0.9168,
+      "step": 21400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4156795476598166e-05,
+      "loss": 0.7983,
+      "step": 21420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4147094155237772e-05,
+      "loss": 0.7842,
+      "step": 21440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4137388117830486e-05,
+      "loss": 0.7559,
+      "step": 21460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4127677375413943e-05,
+      "loss": 0.7416,
+      "step": 21480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4117961939031135e-05,
+      "loss": 0.8395,
+      "step": 21500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4108241819730384e-05,
+      "loss": 0.8557,
+      "step": 21520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.409851702856535e-05,
+      "loss": 0.7085,
+      "step": 21540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4088787576594996e-05,
+      "loss": 0.8655,
+      "step": 21560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4079053474883584e-05,
+      "loss": 0.754,
+      "step": 21580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4069314734500674e-05,
+      "loss": 0.8455,
+      "step": 21600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.405957136652109e-05,
+      "loss": 0.8776,
+      "step": 21620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4049823382024922e-05,
+      "loss": 0.7825,
+      "step": 21640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4040070792097515e-05,
+      "loss": 0.7666,
+      "step": 21660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4030313607829444e-05,
+      "loss": 0.8492,
+      "step": 21680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4020551840316515e-05,
+      "loss": 0.8367,
+      "step": 21700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4010785500659735e-05,
+      "loss": 0.8087,
+      "step": 21720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4001014599965327e-05,
+      "loss": 0.7934,
+      "step": 21740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3991239149344686e-05,
+      "loss": 0.7869,
+      "step": 21760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3981459159914392e-05,
+      "loss": 0.8105,
+      "step": 21780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3971674642796178e-05,
+      "loss": 0.878,
+      "step": 21800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.396188560911693e-05,
+      "loss": 0.8064,
+      "step": 21820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3952092070008669e-05,
+      "loss": 0.8331,
+      "step": 21840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3942294036608544e-05,
+      "loss": 0.8226,
+      "step": 21860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3932491520058807e-05,
+      "loss": 0.797,
+      "step": 21880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.392268453150681e-05,
+      "loss": 0.8333,
+      "step": 21900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3912873082105002e-05,
+      "loss": 0.8006,
+      "step": 21920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3903057183010887e-05,
+      "loss": 0.7301,
+      "step": 21940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3893236845387041e-05,
+      "loss": 0.7378,
+      "step": 21960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3883412080401083e-05,
+      "loss": 0.7564,
+      "step": 21980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3873582899225672e-05,
+      "loss": 0.8053,
+      "step": 22000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3863749313038484e-05,
+      "loss": 0.8445,
+      "step": 22020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3853911333022199e-05,
+      "loss": 0.767,
+      "step": 22040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3844068970364506e-05,
+      "loss": 0.7772,
+      "step": 22060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.383422223625807e-05,
+      "loss": 0.851,
+      "step": 22080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3824371141900533e-05,
+      "loss": 0.76,
+      "step": 22100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3814515698494484e-05,
+      "loss": 0.7725,
+      "step": 22120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3804655917247471e-05,
+      "loss": 0.7981,
+      "step": 22140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3794791809371971e-05,
+      "loss": 0.7698,
+      "step": 22160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3784923386085373e-05,
+      "loss": 0.8396,
+      "step": 22180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3775050658609988e-05,
+      "loss": 0.8182,
+      "step": 22200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3765173638173004e-05,
+      "loss": 0.8189,
+      "step": 22220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3755292336006512e-05,
+      "loss": 0.7969,
+      "step": 22240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3745406763347446e-05,
+      "loss": 0.7591,
+      "step": 22260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3735516931437625e-05,
+      "loss": 0.8119,
+      "step": 22280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3725622851523692e-05,
+      "loss": 0.8237,
+      "step": 22300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3715724534857127e-05,
+      "loss": 0.7985,
+      "step": 22320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3705821992694228e-05,
+      "loss": 0.7909,
+      "step": 22340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3695915236296095e-05,
+      "loss": 0.7241,
+      "step": 22360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3686004276928627e-05,
+      "loss": 0.7716,
+      "step": 22380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3676089125862495e-05,
+      "loss": 0.8163,
+      "step": 22400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3666169794373141e-05,
+      "loss": 0.7784,
+      "step": 22420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3656246293740767e-05,
+      "loss": 0.7793,
+      "step": 22440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.36463186352503e-05,
+      "loss": 0.8619,
+      "step": 22460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3636386830191409e-05,
+      "loss": 0.7603,
+      "step": 22480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3626450889858476e-05,
+      "loss": 0.786,
+      "step": 22500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3616510825550582e-05,
+      "loss": 0.8596,
+      "step": 22520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.36065666485715e-05,
+      "loss": 0.8392,
+      "step": 22540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3596618370229679e-05,
+      "loss": 0.699,
+      "step": 22560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3586666001838231e-05,
+      "loss": 0.7612,
+      "step": 22580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3576709554714923e-05,
+      "loss": 0.8525,
+      "step": 22600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.356674904018216e-05,
+      "loss": 0.8214,
+      "step": 22620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3556784469566968e-05,
+      "loss": 0.7399,
+      "step": 22640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3546815854200987e-05,
+      "loss": 0.8157,
+      "step": 22660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.353684320542046e-05,
+      "loss": 0.8849,
+      "step": 22680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3526866534566213e-05,
+      "loss": 0.8243,
+      "step": 22700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3516885852983645e-05,
+      "loss": 0.7794,
+      "step": 22720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3506901172022724e-05,
+      "loss": 0.7773,
+      "step": 22740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3496912503037959e-05,
+      "loss": 0.7399,
+      "step": 22760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3486919857388392e-05,
+      "loss": 0.8805,
+      "step": 22780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.347692324643759e-05,
+      "loss": 0.7985,
+      "step": 22800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3466922681553636e-05,
+      "loss": 0.7705,
+      "step": 22820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3456918174109102e-05,
+      "loss": 0.8026,
+      "step": 22840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.344690973548104e-05,
+      "loss": 0.7575,
+      "step": 22860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3436897377050982e-05,
+      "loss": 0.7957,
+      "step": 22880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.342688111020491e-05,
+      "loss": 0.7325,
+      "step": 22900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3416860946333256e-05,
+      "loss": 0.8267,
+      "step": 22920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3406836896830884e-05,
+      "loss": 0.8662,
+      "step": 22940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3396808973097066e-05,
+      "loss": 0.8306,
+      "step": 22960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3386777186535495e-05,
+      "loss": 0.8397,
+      "step": 22980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3376741548554248e-05,
+      "loss": 0.7559,
+      "step": 23000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3366702070565775e-05,
+      "loss": 0.801,
+      "step": 23020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3356658763986919e-05,
+      "loss": 0.7433,
+      "step": 23040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.334661164023884e-05,
+      "loss": 0.8088,
+      "step": 23060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3336560710747073e-05,
+      "loss": 0.8296,
+      "step": 23080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3326505986941456e-05,
+      "loss": 0.7567,
+      "step": 23100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3316447480256153e-05,
+      "loss": 0.7866,
+      "step": 23120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3306385202129633e-05,
+      "loss": 0.7891,
+      "step": 23140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3296319164004646e-05,
+      "loss": 0.7994,
+      "step": 23160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3286249377328222e-05,
+      "loss": 0.7251,
+      "step": 23180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.327617585355165e-05,
+      "loss": 0.7805,
+      "step": 23200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3266098604130477e-05,
+      "loss": 0.7611,
+      "step": 23220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.325601764052448e-05,
+      "loss": 0.8033,
+      "step": 23240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.324593297419766e-05,
+      "loss": 0.8043,
+      "step": 23260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.323584461661823e-05,
+      "loss": 0.7901,
+      "step": 23280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3225752579258599e-05,
+      "loss": 0.797,
+      "step": 23300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3215656873595366e-05,
+      "loss": 0.8191,
+      "step": 23320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3205557511109298e-05,
+      "loss": 0.7936,
+      "step": 23340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.319545450328532e-05,
+      "loss": 0.8261,
+      "step": 23360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3185347861612501e-05,
+      "loss": 0.7741,
+      "step": 23380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3175237597584045e-05,
+      "loss": 0.7602,
+      "step": 23400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3165123722697276e-05,
+      "loss": 0.7664,
+      "step": 23420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3155006248453615e-05,
+      "loss": 0.7819,
+      "step": 23440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3144885186358596e-05,
+      "loss": 0.7912,
+      "step": 23460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3134760547921814e-05,
+      "loss": 0.8161,
+      "step": 23480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3124632344656943e-05,
+      "loss": 0.7957,
+      "step": 23500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3114500588081698e-05,
+      "loss": 0.8164,
+      "step": 23520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3104365289717849e-05,
+      "loss": 0.7976,
+      "step": 23540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3094226461091186e-05,
+      "loss": 0.7437,
+      "step": 23560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3084084113731514e-05,
+      "loss": 0.84,
+      "step": 23580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3073938259172645e-05,
+      "loss": 0.7651,
+      "step": 23600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3063788908952368e-05,
+      "loss": 0.7769,
+      "step": 23620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3053636074612457e-05,
+      "loss": 0.7899,
+      "step": 23640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3043479767698646e-05,
+      "loss": 0.7696,
+      "step": 23660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3033319999760619e-05,
+      "loss": 0.7587,
+      "step": 23680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3023156782351989e-05,
+      "loss": 0.7693,
+      "step": 23700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.30129901270303e-05,
+      "loss": 0.7695,
+      "step": 23720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3002820045357005e-05,
+      "loss": 0.79,
+      "step": 23740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2992646548897442e-05,
+      "loss": 0.8037,
+      "step": 23760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2982469649220854e-05,
+      "loss": 0.8023,
+      "step": 23780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2972289357900332e-05,
+      "loss": 0.8247,
+      "step": 23800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2962105686512834e-05,
+      "loss": 0.8243,
+      "step": 23820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2951918646639163e-05,
+      "loss": 0.8695,
+      "step": 23840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.294172824986395e-05,
+      "loss": 0.8408,
+      "step": 23860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.293153450777564e-05,
+      "loss": 0.8392,
+      "step": 23880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2921337431966494e-05,
+      "loss": 0.7882,
+      "step": 23900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2911137034032553e-05,
+      "loss": 0.8327,
+      "step": 23920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2900933325573636e-05,
+      "loss": 0.8817,
+      "step": 23940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2890726318193332e-05,
+      "loss": 0.8262,
+      "step": 23960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2880516023498981e-05,
+      "loss": 0.7659,
+      "step": 23980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2870302453101657e-05,
+      "loss": 0.7805,
+      "step": 24000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2860085618616169e-05,
+      "loss": 0.8152,
+      "step": 24020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.284986553166102e-05,
+      "loss": 0.7744,
+      "step": 24040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2839642203858433e-05,
+      "loss": 0.7241,
+      "step": 24060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2829415646834295e-05,
+      "loss": 0.8229,
+      "step": 24080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2819185872218184e-05,
+      "loss": 0.8712,
+      "step": 24100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2808952891643327e-05,
+      "loss": 0.7865,
+      "step": 24120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2798716716746602e-05,
+      "loss": 0.8225,
+      "step": 24140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2788477359168512e-05,
+      "loss": 0.7767,
+      "step": 24160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2778234830553183e-05,
+      "loss": 0.8087,
+      "step": 24180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.276798914254835e-05,
+      "loss": 0.8498,
+      "step": 24200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2757740306805345e-05,
+      "loss": 0.785,
+      "step": 24220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2747488334979064e-05,
+      "loss": 0.7809,
+      "step": 24240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2737233238727982e-05,
+      "loss": 0.77,
+      "step": 24260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2726975029714124e-05,
+      "loss": 0.8828,
+      "step": 24280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.2716713719603053e-05,
+      "loss": 0.8708,
+      "step": 24300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2706449320063866e-05,
+      "loss": 0.8074,
+      "step": 24320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2696181842769161e-05,
+      "loss": 0.7782,
+      "step": 24340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2685911299395046e-05,
+      "loss": 0.8194,
+      "step": 24360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2675637701621111e-05,
+      "loss": 0.7642,
+      "step": 24380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2665361061130416e-05,
+      "loss": 0.7517,
+      "step": 24400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2655081389609496e-05,
+      "loss": 0.7556,
+      "step": 24420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2644798698748314e-05,
+      "loss": 0.7667,
+      "step": 24440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2634513000240277e-05,
+      "loss": 0.7696,
+      "step": 24460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2624224305782215e-05,
+      "loss": 0.8711,
+      "step": 24480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2613932627074356e-05,
+      "loss": 0.7339,
+      "step": 24500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2603637975820324e-05,
+      "loss": 0.7713,
+      "step": 24520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2593340363727131e-05,
+      "loss": 0.7791,
+      "step": 24540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2583039802505148e-05,
+      "loss": 0.8241,
+      "step": 24560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.2572736303868102e-05,
+      "loss": 0.7267,
+      "step": 24580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.256242987953306e-05,
+      "loss": 0.8083,
+      "step": 24600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2552120541220419e-05,
+      "loss": 0.8438,
+      "step": 24620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2541808300653883e-05,
+      "loss": 0.7614,
+      "step": 24640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.253149316956047e-05,
+      "loss": 0.7978,
+      "step": 24660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2521175159670464e-05,
+      "loss": 0.7852,
+      "step": 24680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2510854282717444e-05,
+      "loss": 0.7439,
+      "step": 24700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2500530550438232e-05,
+      "loss": 0.8229,
+      "step": 24720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2490203974572915e-05,
+      "loss": 0.7607,
+      "step": 24740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2479874566864799e-05,
+      "loss": 0.7557,
+      "step": 24760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2469542339060414e-05,
+      "loss": 0.7902,
+      "step": 24780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2459207302909499e-05,
+      "loss": 0.7407,
+      "step": 24800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2448869470164988e-05,
+      "loss": 0.8128,
+      "step": 24820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.243852885258299e-05,
+      "loss": 0.7656,
+      "step": 24840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2428185461922783e-05,
+      "loss": 0.7291,
+      "step": 24860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2417839309946806e-05,
+      "loss": 0.844,
+      "step": 24880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2407490408420629e-05,
+      "loss": 0.7925,
+      "step": 24900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2397138769112949e-05,
+      "loss": 0.7518,
+      "step": 24920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2386784403795577e-05,
+      "loss": 0.7826,
+      "step": 24940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2376427324243432e-05,
+      "loss": 0.8126,
+      "step": 24960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2366067542234512e-05,
+      "loss": 0.7521,
+      "step": 24980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2355705069549885e-05,
+      "loss": 0.8282,
+      "step": 25000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2345339917973688e-05,
+      "loss": 0.7748,
+      "step": 25020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2334972099293102e-05,
+      "loss": 0.8185,
+      "step": 25040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2324601625298331e-05,
+      "loss": 0.8449,
+      "step": 25060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2314228507782614e-05,
+      "loss": 0.8139,
+      "step": 25080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2303852758542187e-05,
+      "loss": 0.7789,
+      "step": 25100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2293474389376281e-05,
+      "loss": 0.8105,
+      "step": 25120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2283093412087099e-05,
+      "loss": 0.7276,
+      "step": 25140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2272709838479824e-05,
+      "loss": 0.8363,
+      "step": 25160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2262323680362587e-05,
+      "loss": 0.7583,
+      "step": 25180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2251934949546446e-05,
+      "loss": 0.7626,
+      "step": 25200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2241543657845403e-05,
+      "loss": 0.7718,
+      "step": 25220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2231149817076358e-05,
+      "loss": 0.7381,
+      "step": 25240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2220753439059116e-05,
+      "loss": 0.8,
+      "step": 25260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2210354535616366e-05,
+      "loss": 0.7758,
+      "step": 25280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2199953118573674e-05,
+      "loss": 0.7634,
+      "step": 25300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2189549199759454e-05,
+      "loss": 0.7878,
+      "step": 25320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2179142791004974e-05,
+      "loss": 0.7906,
+      "step": 25340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2168733904144328e-05,
+      "loss": 0.7498,
+      "step": 25360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2158322551014431e-05,
+      "loss": 0.7985,
+      "step": 25380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2147908743455005e-05,
+      "loss": 0.7641,
+      "step": 25400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2137492493308558e-05,
+      "loss": 0.7765,
+      "step": 25420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2127073812420376e-05,
+      "loss": 0.81,
+      "step": 25440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2116652712638513e-05,
+      "loss": 0.7686,
+      "step": 25460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.210622920581377e-05,
+      "loss": 0.7756,
+      "step": 25480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2095803303799693e-05,
+      "loss": 0.8119,
+      "step": 25500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2085375018452541e-05,
+      "loss": 0.6548,
+      "step": 25520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2074944361631288e-05,
+      "loss": 0.7717,
+      "step": 25540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2064511345197606e-05,
+      "loss": 0.8245,
+      "step": 25560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.205407598101585e-05,
+      "loss": 0.7411,
+      "step": 25580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2043638280953043e-05,
+      "loss": 0.7405,
+      "step": 25600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2033198256878867e-05,
+      "loss": 0.7835,
+      "step": 25620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2022755920665648e-05,
+      "loss": 0.7589,
+      "step": 25640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.201231128418833e-05,
+      "loss": 0.7958,
+      "step": 25660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.200186435932449e-05,
+      "loss": 0.7981,
+      "step": 25680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1991415157954293e-05,
+      "loss": 0.776,
+      "step": 25700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1980963691960508e-05,
+      "loss": 0.7557,
+      "step": 25720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1970509973228459e-05,
+      "loss": 0.803,
+      "step": 25740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1960054013646048e-05,
+      "loss": 0.7916,
+      "step": 25760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1949595825103721e-05,
+      "loss": 0.766,
+      "step": 25780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1939135419494456e-05,
+      "loss": 0.7568,
+      "step": 25800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.192867280871376e-05,
+      "loss": 0.7633,
+      "step": 25820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1918208004659632e-05,
+      "loss": 0.7924,
+      "step": 25840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1907741019232584e-05,
+      "loss": 0.7998,
+      "step": 25860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1897271864335594e-05,
+      "loss": 0.7736,
+      "step": 25880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1886800551874111e-05,
+      "loss": 0.7891,
+      "step": 25900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1876327093756048e-05,
+      "loss": 0.72,
+      "step": 25920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1865851501891741e-05,
+      "loss": 0.753,
+      "step": 25940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1855373788193969e-05,
+      "loss": 0.7819,
+      "step": 25960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1844893964577906e-05,
+      "loss": 0.7506,
+      "step": 25980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1834412042961143e-05,
+      "loss": 0.7076,
+      "step": 26000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1823928035263646e-05,
+      "loss": 0.7446,
+      "step": 26020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1813441953407754e-05,
+      "loss": 0.7617,
+      "step": 26040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1802953809318173e-05,
+      "loss": 0.783,
+      "step": 26060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1792463614921945e-05,
+      "loss": 0.7408,
+      "step": 26080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1781971382148446e-05,
+      "loss": 0.7503,
+      "step": 26100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.177147712292937e-05,
+      "loss": 0.8042,
+      "step": 26120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1760980849198717e-05,
+      "loss": 0.7961,
+      "step": 26140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1750482572892781e-05,
+      "loss": 0.75,
+      "step": 26160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1739982305950124e-05,
+      "loss": 0.7585,
+      "step": 26180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.172948006031158e-05,
+      "loss": 0.7957,
+      "step": 26200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1718975847920229e-05,
+      "loss": 0.7645,
+      "step": 26220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.170846968072139e-05,
+      "loss": 0.7516,
+      "step": 26240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1697961570662606e-05,
+      "loss": 0.7382,
+      "step": 26260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1687451529693626e-05,
+      "loss": 0.834,
+      "step": 26280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.167693956976639e-05,
+      "loss": 0.8297,
+      "step": 26300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1666425702835033e-05,
+      "loss": 0.7982,
+      "step": 26320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1655909940855852e-05,
+      "loss": 0.7996,
+      "step": 26340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1645392295787299e-05,
+      "loss": 0.7665,
+      "step": 26360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1634872779589966e-05,
+      "loss": 0.824,
+      "step": 26380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1624351404226573e-05,
+      "loss": 0.7679,
+      "step": 26400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1613828181661957e-05,
+      "loss": 0.8051,
+      "step": 26420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1603303123863054e-05,
+      "loss": 0.7372,
+      "step": 26440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1592776242798887e-05,
+      "loss": 0.7109,
+      "step": 26460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1582247550440555e-05,
+      "loss": 0.7646,
+      "step": 26480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1571717058761211e-05,
+      "loss": 0.7508,
+      "step": 26500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1561184779736061e-05,
+      "loss": 0.7319,
+      "step": 26520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1550650725342337e-05,
+      "loss": 0.6879,
+      "step": 26540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1540114907559294e-05,
+      "loss": 0.7612,
+      "step": 26560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1529577338368191e-05,
+      "loss": 0.7385,
+      "step": 26580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1519038029752281e-05,
+      "loss": 0.7635,
+      "step": 26600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1508496993696791e-05,
+      "loss": 0.6925,
+      "step": 26620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1497954242188913e-05,
+      "loss": 0.7441,
+      "step": 26640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.14874097872178e-05,
+      "loss": 0.7624,
+      "step": 26660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1476863640774523e-05,
+      "loss": 0.7689,
+      "step": 26680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1466315814852094e-05,
+      "loss": 0.7357,
+      "step": 26700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.145576632144542e-05,
+      "loss": 0.7586,
+      "step": 26720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.144521517255132e-05,
+      "loss": 0.7609,
+      "step": 26740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1434662380168487e-05,
+      "loss": 0.7013,
+      "step": 26760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1424107956297479e-05,
+      "loss": 0.7781,
+      "step": 26780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1413551912940721e-05,
+      "loss": 0.7621,
+      "step": 26800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1402994262102468e-05,
+      "loss": 0.7476,
+      "step": 26820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1392435015788805e-05,
+      "loss": 0.6879,
+      "step": 26840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.138187418600764e-05,
+      "loss": 0.7534,
+      "step": 26860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1371311784768672e-05,
+      "loss": 0.8091,
+      "step": 26880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1360747824083397e-05,
+      "loss": 0.743,
+      "step": 26900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1350182315965071e-05,
+      "loss": 0.8519,
+      "step": 26920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1339615272428722e-05,
+      "loss": 0.7123,
+      "step": 26940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1329046705491111e-05,
+      "loss": 0.7269,
+      "step": 26960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1318476627170751e-05,
+      "loss": 0.8065,
+      "step": 26980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1307905049487855e-05,
+      "loss": 0.7649,
+      "step": 27000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.129733198446435e-05,
+      "loss": 0.7449,
+      "step": 27020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1286757444123854e-05,
+      "loss": 0.761,
+      "step": 27040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1276181440491658e-05,
+      "loss": 0.7445,
+      "step": 27060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1265603985594723e-05,
+      "loss": 0.758,
+      "step": 27080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1255025091461657e-05,
+      "loss": 0.7528,
+      "step": 27100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1244444770122707e-05,
+      "loss": 0.7695,
+      "step": 27120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.123386303360974e-05,
+      "loss": 0.7742,
+      "step": 27140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1223279893956234e-05,
+      "loss": 0.8097,
+      "step": 27160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1212695363197262e-05,
+      "loss": 0.8056,
+      "step": 27180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.120210945336948e-05,
+      "loss": 0.7472,
+      "step": 27200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1191522176511112e-05,
+      "loss": 0.8154,
+      "step": 27220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1180933544661936e-05,
+      "loss": 0.736,
+      "step": 27240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1170343569863274e-05,
+      "loss": 0.7416,
+      "step": 27260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1159752264157963e-05,
+      "loss": 0.768,
+      "step": 27280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1149159639590377e-05,
+      "loss": 0.7803,
+      "step": 27300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1138565708206364e-05,
+      "loss": 0.7374,
+      "step": 27320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1127970482053274e-05,
+      "loss": 0.7396,
+      "step": 27340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1117373973179925e-05,
+      "loss": 0.7836,
+      "step": 27360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1106776193636595e-05,
+      "loss": 0.7813,
+      "step": 27380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1096177155475006e-05,
+      "loss": 0.805,
+      "step": 27400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1085576870748307e-05,
+      "loss": 0.7897,
+      "step": 27420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1074975351511075e-05,
+      "loss": 0.7369,
+      "step": 27440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1064372609819279e-05,
+      "loss": 0.7486,
+      "step": 27460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1053768657730284e-05,
+      "loss": 0.7335,
+      "step": 27480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1043163507302835e-05,
+      "loss": 0.8417,
+      "step": 27500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1032557170597034e-05,
+      "loss": 0.6774,
+      "step": 27520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1021949659674337e-05,
+      "loss": 0.7958,
+      "step": 27540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1011340986597528e-05,
+      "loss": 0.8394,
+      "step": 27560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.100073116343072e-05,
+      "loss": 0.799,
+      "step": 27580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0990120202239324e-05,
+      "loss": 0.8001,
+      "step": 27600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.097950811509006e-05,
+      "loss": 0.7543,
+      "step": 27620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.096889491405092e-05,
+      "loss": 0.7774,
+      "step": 27640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0958280611191162e-05,
+      "loss": 0.7711,
+      "step": 27660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0947665218581293e-05,
+      "loss": 0.8092,
+      "step": 27680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0937048748293068e-05,
+      "loss": 0.7781,
+      "step": 27700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0926431212399467e-05,
+      "loss": 0.7465,
+      "step": 27720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0915812622974673e-05,
+      "loss": 0.7258,
+      "step": 27740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0905192992094075e-05,
+      "loss": 0.7642,
+      "step": 27760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0894572331834248e-05,
+      "loss": 0.7676,
+      "step": 27780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0883950654272925e-05,
+      "loss": 0.838,
+      "step": 27800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0873327971489015e-05,
+      "loss": 0.7291,
+      "step": 27820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.086270429556255e-05,
+      "loss": 0.7436,
+      "step": 27840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0852079638574708e-05,
+      "loss": 0.7519,
+      "step": 27860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.084145401260777e-05,
+      "loss": 0.8155,
+      "step": 27880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.083082742974513e-05,
+      "loss": 0.7486,
+      "step": 27900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.082019990207126e-05,
+      "loss": 0.6917,
+      "step": 27920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0809571441671716e-05,
+      "loss": 0.7539,
+      "step": 27940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0798942060633108e-05,
+      "loss": 0.7732,
+      "step": 27960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0788311771043092e-05,
+      "loss": 0.7179,
+      "step": 27980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0777680584990366e-05,
+      "loss": 0.8022,
+      "step": 28000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0767048514564633e-05,
+      "loss": 0.8168,
+      "step": 28020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0756415571856617e-05,
+      "loss": 0.766,
+      "step": 28040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.074578176895803e-05,
+      "loss": 0.744,
+      "step": 28060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.073514711796155e-05,
+      "loss": 0.7179,
+      "step": 28080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0724511630960837e-05,
+      "loss": 0.7154,
+      "step": 28100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0713875320050487e-05,
+      "loss": 0.7586,
+      "step": 28120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0703238197326042e-05,
+      "loss": 0.828,
+      "step": 28140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0692600274883969e-05,
+      "loss": 0.7452,
+      "step": 28160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0681961564821632e-05,
+      "loss": 0.7193,
+      "step": 28180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0671322079237307e-05,
+      "loss": 0.7795,
+      "step": 28200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0660681830230137e-05,
+      "loss": 0.7692,
+      "step": 28220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0650040829900143e-05,
+      "loss": 0.7935,
+      "step": 28240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.06393990903482e-05,
+      "loss": 0.7645,
+      "step": 28260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0628756623676016e-05,
+      "loss": 0.8378,
+      "step": 28280
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.061811344198613e-05,
+      "loss": 0.7507,
+      "step": 28300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.06074695573819e-05,
+      "loss": 0.8633,
+      "step": 28320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0596824981967475e-05,
+      "loss": 0.7645,
+      "step": 28340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0586179727847791e-05,
+      "loss": 0.7723,
+      "step": 28360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0575533807128567e-05,
+      "loss": 0.8174,
+      "step": 28380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0564887231916256e-05,
+      "loss": 0.7738,
+      "step": 28400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0554240014318085e-05,
+      "loss": 0.7467,
+      "step": 28420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0543592166441983e-05,
+      "loss": 0.7665,
+      "step": 28440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0532943700396616e-05,
+      "loss": 0.7758,
+      "step": 28460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0522294628291347e-05,
+      "loss": 0.7501,
+      "step": 28480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0511644962236228e-05,
+      "loss": 0.7612,
+      "step": 28500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0500994714341984e-05,
+      "loss": 0.7664,
+      "step": 28520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.049034389672e-05,
+      "loss": 0.7437,
+      "step": 28540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0479692521482316e-05,
+      "loss": 0.718,
+      "step": 28560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0469040600741603e-05,
+      "loss": 0.7596,
+      "step": 28580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0458388146611151e-05,
+      "loss": 0.8108,
+      "step": 28600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.044773517120486e-05,
+      "loss": 0.7887,
+      "step": 28620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0437081686637216e-05,
+      "loss": 0.7195,
+      "step": 28640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0426427705023289e-05,
+      "loss": 0.7793,
+      "step": 28660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0415773238478715e-05,
+      "loss": 0.6975,
+      "step": 28680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0405118299119682e-05,
+      "loss": 0.7211,
+      "step": 28700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.039446289906291e-05,
+      "loss": 0.7757,
+      "step": 28720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0383807050425648e-05,
+      "loss": 0.7514,
+      "step": 28740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.037315076532565e-05,
+      "loss": 0.6998,
+      "step": 28760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0362494055881172e-05,
+      "loss": 0.7226,
+      "step": 28780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0351836934210957e-05,
+      "loss": 0.7645,
+      "step": 28800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.03411794124342e-05,
+      "loss": 0.8352,
+      "step": 28820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0330521502670574e-05,
+      "loss": 0.7759,
+      "step": 28840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0319863217040166e-05,
+      "loss": 0.7611,
+      "step": 28860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0309204567663513e-05,
+      "loss": 0.7064,
+      "step": 28880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0298545566661558e-05,
+      "loss": 0.6901,
+      "step": 28900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.028788622615564e-05,
+      "loss": 0.7307,
+      "step": 28920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0277226558267494e-05,
+      "loss": 0.7325,
+      "step": 28940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0266566575119213e-05,
+      "loss": 0.8501,
+      "step": 28960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0255906288833259e-05,
+      "loss": 0.7255,
+      "step": 28980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0245245711532437e-05,
+      "loss": 0.6879,
+      "step": 29000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0234584855339884e-05,
+      "loss": 0.7236,
+      "step": 29020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0223923732379049e-05,
+      "loss": 0.761,
+      "step": 29040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0213262354773689e-05,
+      "loss": 0.7608,
+      "step": 29060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.020260073464785e-05,
+      "loss": 0.7164,
+      "step": 29080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0191938884125847e-05,
+      "loss": 0.7795,
+      "step": 29100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0181276815332273e-05,
+      "loss": 0.6521,
+      "step": 29120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0170614540391952e-05,
+      "loss": 0.7608,
+      "step": 29140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0159952071429953e-05,
+      "loss": 0.7849,
+      "step": 29160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0149289420571558e-05,
+      "loss": 0.6989,
+      "step": 29180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0138626599942263e-05,
+      "loss": 0.7986,
+      "step": 29200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0127963621667754e-05,
+      "loss": 0.7628,
+      "step": 29220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0117300497873897e-05,
+      "loss": 0.7438,
+      "step": 29240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0106637240686723e-05,
+      "loss": 0.7998,
+      "step": 29260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.009597386223241e-05,
+      "loss": 0.7247,
+      "step": 29280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0085310374637288e-05,
+      "loss": 0.7975,
+      "step": 29300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0074646790027792e-05,
+      "loss": 0.7294,
+      "step": 29320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.006398312053048e-05,
+      "loss": 0.7467,
+      "step": 29340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0053319378272008e-05,
+      "loss": 0.7376,
+      "step": 29360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0042655575379103e-05,
+      "loss": 0.8358,
+      "step": 29380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0031991723978574e-05,
+      "loss": 0.8395,
+      "step": 29400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0021327836197274e-05,
+      "loss": 0.7004,
+      "step": 29420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.001066392416211e-05,
+      "loss": 0.6753,
+      "step": 29440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1e-05,
+      "loss": 0.7482,
+      "step": 29460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.989336075837893e-06,
+      "loss": 0.5971,
+      "step": 29480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.978672163802726e-06,
+      "loss": 0.5315,
+      "step": 29500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.96800827602143e-06,
+      "loss": 0.5765,
+      "step": 29520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.9573444246209e-06,
+      "loss": 0.5696,
+      "step": 29540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.946680621727995e-06,
+      "loss": 0.5824,
+      "step": 29560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.936016879469522e-06,
+      "loss": 0.5709,
+      "step": 29580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.925353209972213e-06,
+      "loss": 0.5152,
+      "step": 29600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.914689625362717e-06,
+      "loss": 0.5453,
+      "step": 29620
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.904026137767591e-06,
+      "loss": 0.5355,
+      "step": 29640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.89336275931328e-06,
+      "loss": 0.5089,
+      "step": 29660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.882699502126103e-06,
+      "loss": 0.5548,
+      "step": 29680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.872036378332251e-06,
+      "loss": 0.5285,
+      "step": 29700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.86137340005774e-06,
+      "loss": 0.5213,
+      "step": 29720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.850710579428445e-06,
+      "loss": 0.4975,
+      "step": 29740
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.84004792857005e-06,
+      "loss": 0.5456,
+      "step": 29760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.82938545960805e-06,
+      "loss": 0.5541,
+      "step": 29780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.818723184667732e-06,
+      "loss": 0.5548,
+      "step": 29800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.808061115874156e-06,
+      "loss": 0.5818,
+      "step": 29820
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.797399265352154e-06,
+      "loss": 0.5393,
+      "step": 29840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.786737645226313e-06,
+      "loss": 0.5183,
+      "step": 29860
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.776076267620955e-06,
+      "loss": 0.5039,
+      "step": 29880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.76541514466012e-06,
+      "loss": 0.5041,
+      "step": 29900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.754754288467564e-06,
+      "loss": 0.4768,
+      "step": 29920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.744093711166745e-06,
+      "loss": 0.5311,
+      "step": 29940
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.733433424880789e-06,
+      "loss": 0.5408,
+      "step": 29960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.72277344173251e-06,
+      "loss": 0.4999,
+      "step": 29980
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.712113773844361e-06,
+      "loss": 0.5325,
+      "step": 30000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.701454433338445e-06,
+      "loss": 0.5887,
+      "step": 30020
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.690795432336489e-06,
+      "loss": 0.5308,
+      "step": 30040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.680136782959835e-06,
+      "loss": 0.5197,
+      "step": 30060
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.669478497329431e-06,
+      "loss": 0.58,
+      "step": 30080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.658820587565801e-06,
+      "loss": 0.5552,
+      "step": 30100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.648163065789046e-06,
+      "loss": 0.5623,
+      "step": 30120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.637505944118828e-06,
+      "loss": 0.5321,
+      "step": 30140
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.626849234674356e-06,
+      "loss": 0.5571,
+      "step": 30160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.616192949574357e-06,
+      "loss": 0.5319,
+      "step": 30180
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.605537100937094e-06,
+      "loss": 0.5443,
+      "step": 30200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.59488170088032e-06,
+      "loss": 0.5153,
+      "step": 30220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.584226761521285e-06,
+      "loss": 0.5195,
+      "step": 30240
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.573572294976714e-06,
+      "loss": 0.5135,
+      "step": 30260
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.562918313362787e-06,
+      "loss": 0.5177,
+      "step": 30280
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.552264828795142e-06,
+      "loss": 0.5544,
+      "step": 30300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.541611853388849e-06,
+      "loss": 0.5243,
+      "step": 30320
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.530959399258397e-06,
+      "loss": 0.5352,
+      "step": 30340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.520307478517686e-06,
+      "loss": 0.5354,
+      "step": 30360
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.509656103280004e-06,
+      "loss": 0.5097,
+      "step": 30380
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.499005285658021e-06,
+      "loss": 0.5286,
+      "step": 30400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.488355037763774e-06,
+      "loss": 0.5257,
+      "step": 30420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.477705371708657e-06,
+      "loss": 0.5127,
+      "step": 30440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.467056299603388e-06,
+      "loss": 0.542,
+      "step": 30460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.456407833558019e-06,
+      "loss": 0.5338,
+      "step": 30480
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.445759985681919e-06,
+      "loss": 0.5501,
+      "step": 30500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.435112768083742e-06,
+      "loss": 0.5282,
+      "step": 30520
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.42446619287144e-06,
+      "loss": 0.5156,
+      "step": 30540
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.41382027215221e-06,
+      "loss": 0.534,
+      "step": 30560
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.403175018032528e-06,
+      "loss": 0.5237,
+      "step": 30580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.3925304426181e-06,
+      "loss": 0.5067,
+      "step": 30600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.38188655801387e-06,
+      "loss": 0.5491,
+      "step": 30620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.37124337632399e-06,
+      "loss": 0.5151,
+      "step": 30640
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.360600909651804e-06,
+      "loss": 0.5706,
+      "step": 30660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.349959170099859e-06,
+      "loss": 0.4919,
+      "step": 30680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.339318169769865e-06,
+      "loss": 0.5317,
+      "step": 30700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.328677920762698e-06,
+      "loss": 0.5356,
+      "step": 30720
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.318038435178371e-06,
+      "loss": 0.574,
+      "step": 30740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.307399725116035e-06,
+      "loss": 0.5253,
+      "step": 30760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.29676180267396e-06,
+      "loss": 0.5226,
+      "step": 30780
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.286124679949515e-06,
+      "loss": 0.5003,
+      "step": 30800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.275488369039169e-06,
+      "loss": 0.569,
+      "step": 30820
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.264852882038453e-06,
+      "loss": 0.5178,
+      "step": 30840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.254218231041973e-06,
+      "loss": 0.519,
+      "step": 30860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.243584428143383e-06,
+      "loss": 0.5543,
+      "step": 30880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.232951485435367e-06,
+      "loss": 0.5391,
+      "step": 30900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.222319415009639e-06,
+      "loss": 0.51,
+      "step": 30920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.21168822895691e-06,
+      "loss": 0.5161,
+      "step": 30940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.201057939366896e-06,
+      "loss": 0.5187,
+      "step": 30960
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.190428558328285e-06,
+      "loss": 0.4999,
+      "step": 30980
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.179800097928743e-06,
+      "loss": 0.4937,
+      "step": 31000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.169172570254874e-06,
+      "loss": 0.5363,
+      "step": 31020
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.158545987392233e-06,
+      "loss": 0.4939,
+      "step": 31040
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.147920361425295e-06,
+      "loss": 0.5205,
+      "step": 31060
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.13729570443745e-06,
+      "loss": 0.5278,
+      "step": 31080
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.12667202851099e-06,
+      "loss": 0.5109,
+      "step": 31100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.116049345727076e-06,
+      "loss": 0.5284,
+      "step": 31120
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.105427668165755e-06,
+      "loss": 0.549,
+      "step": 31140
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.094807007905927e-06,
+      "loss": 0.5686,
+      "step": 31160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.084187377025334e-06,
+      "loss": 0.5356,
+      "step": 31180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.073568787600538e-06,
+      "loss": 0.4808,
+      "step": 31200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.062951251706933e-06,
+      "loss": 0.5262,
+      "step": 31220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.052334781418708e-06,
+      "loss": 0.5427,
+      "step": 31240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.04171938880884e-06,
+      "loss": 0.5331,
+      "step": 31260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.031105085949083e-06,
+      "loss": 0.501,
+      "step": 31280
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.020491884909941e-06,
+      "loss": 0.5483,
+      "step": 31300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.009879797760679e-06,
+      "loss": 0.5403,
+      "step": 31320
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.999268836569284e-06,
+      "loss": 0.5151,
+      "step": 31340
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.988659013402474e-06,
+      "loss": 0.5153,
+      "step": 31360
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.978050340325668e-06,
+      "loss": 0.5419,
+      "step": 31380
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.967442829402968e-06,
+      "loss": 0.4932,
+      "step": 31400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.956836492697167e-06,
+      "loss": 0.5351,
+      "step": 31420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.94623134226972e-06,
+      "loss": 0.5007,
+      "step": 31440
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.935627390180728e-06,
+      "loss": 0.5453,
+      "step": 31460
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.925024648488932e-06,
+      "loss": 0.528,
+      "step": 31480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.914423129251697e-06,
+      "loss": 0.5672,
+      "step": 31500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.903822844524999e-06,
+      "loss": 0.5049,
+      "step": 31520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.893223806363407e-06,
+      "loss": 0.5704,
+      "step": 31540
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.882626026820078e-06,
+      "loss": 0.4887,
+      "step": 31560
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.872029517946729e-06,
+      "loss": 0.5191,
+      "step": 31580
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.861434291793639e-06,
+      "loss": 0.5468,
+      "step": 31600
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.850840360409627e-06,
+      "loss": 0.5635,
+      "step": 31620
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.840247735842037e-06,
+      "loss": 0.5354,
+      "step": 31640
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.829656430136732e-06,
+      "loss": 0.4913,
+      "step": 31660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.819066455338067e-06,
+      "loss": 0.5314,
+      "step": 31680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.80847782348889e-06,
+      "loss": 0.5026,
+      "step": 31700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.797890546630522e-06,
+      "loss": 0.5835,
+      "step": 31720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.787304636802743e-06,
+      "loss": 0.5104,
+      "step": 31740
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.77672010604377e-06,
+      "loss": 0.4618,
+      "step": 31760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.766136966390263e-06,
+      "loss": 0.5285,
+      "step": 31780
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.755555229877295e-06,
+      "loss": 0.5043,
+      "step": 31800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.744974908538345e-06,
+      "loss": 0.5348,
+      "step": 31820
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.734396014405282e-06,
+      "loss": 0.5492,
+      "step": 31840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.723818559508345e-06,
+      "loss": 0.5543,
+      "step": 31860
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.713242555876149e-06,
+      "loss": 0.5509,
+      "step": 31880
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.702668015535651e-06,
+      "loss": 0.5401,
+      "step": 31900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.692094950512145e-06,
+      "loss": 0.5644,
+      "step": 31920
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.681523372829254e-06,
+      "loss": 0.5083,
+      "step": 31940
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.670953294508892e-06,
+      "loss": 0.5296,
+      "step": 31960
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.660384727571282e-06,
+      "loss": 0.6039,
+      "step": 31980
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.64981768403493e-06,
+      "loss": 0.4802,
+      "step": 32000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.639252175916608e-06,
+      "loss": 0.5125,
+      "step": 32020
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.62868821523133e-06,
+      "loss": 0.4873,
+      "step": 32040
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.618125813992364e-06,
+      "loss": 0.53,
+      "step": 32060
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.607564984211199e-06,
+      "loss": 0.5251,
+      "step": 32080
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.597005737897534e-06,
+      "loss": 0.57,
+      "step": 32100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.586448087059284e-06,
+      "loss": 0.5491,
+      "step": 32120
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.575892043702523e-06,
+      "loss": 0.5388,
+      "step": 32140
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.565337619831516e-06,
+      "loss": 0.5428,
+      "step": 32160
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.554784827448681e-06,
+      "loss": 0.5715,
+      "step": 32180
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.544233678554581e-06,
+      "loss": 0.5013,
+      "step": 32200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.533684185147913e-06,
+      "loss": 0.5551,
+      "step": 32220
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.52313635922548e-06,
+      "loss": 0.5181,
+      "step": 32240
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.512590212782203e-06,
+      "loss": 0.5321,
+      "step": 32260
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.502045757811087e-06,
+      "loss": 0.5279,
+      "step": 32280
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.491503006303212e-06,
+      "loss": 0.5403,
+      "step": 32300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.480961970247722e-06,
+      "loss": 0.5359,
+      "step": 32320
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.47042266163181e-06,
+      "loss": 0.533,
+      "step": 32340
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.459885092440708e-06,
+      "loss": 0.5727,
+      "step": 32360
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.449349274657665e-06,
+      "loss": 0.5204,
+      "step": 32380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.438815220263942e-06,
+      "loss": 0.5187,
+      "step": 32400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.42828294123879e-06,
+      "loss": 0.5268,
+      "step": 32420
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.417752449559447e-06,
+      "loss": 0.5333,
+      "step": 32440
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.407223757201113e-06,
+      "loss": 0.5326,
+      "step": 32460
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.396696876136948e-06,
+      "loss": 0.5232,
+      "step": 32480
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.386171818338048e-06,
+      "loss": 0.4898,
+      "step": 32500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.37564859577343e-06,
+      "loss": 0.4699,
+      "step": 32520
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.365127220410037e-06,
+      "loss": 0.5222,
+      "step": 32540
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.354607704212703e-06,
+      "loss": 0.5207,
+      "step": 32560
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.344090059144151e-06,
+      "loss": 0.5525,
+      "step": 32580
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.333574297164969e-06,
+      "loss": 0.5413,
+      "step": 32600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.323060430233613e-06,
+      "loss": 0.5338,
+      "step": 32620
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.312548470306379e-06,
+      "loss": 0.535,
+      "step": 32640
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.302038429337396e-06,
+      "loss": 0.5453,
+      "step": 32660
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.291530319278614e-06,
+      "loss": 0.5435,
+      "step": 32680
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.281024152079775e-06,
+      "loss": 0.5129,
+      "step": 32700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.270519939688421e-06,
+      "loss": 0.5429,
+      "step": 32720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.260017694049878e-06,
+      "loss": 0.5661,
+      "step": 32740
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.249517427107225e-06,
+      "loss": 0.5217,
+      "step": 32760
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.239019150801286e-06,
+      "loss": 0.4945,
+      "step": 32780
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.228522877070636e-06,
+      "loss": 0.5153,
+      "step": 32800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.21802861785156e-06,
+      "loss": 0.5137,
+      "step": 32820
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.207536385078057e-06,
+      "loss": 0.5012,
+      "step": 32840
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.197046190681832e-06,
+      "loss": 0.5243,
+      "step": 32860
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.186558046592247e-06,
+      "loss": 0.5136,
+      "step": 32880
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.176071964736358e-06,
+      "loss": 0.4767,
+      "step": 32900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.16558795703886e-06,
+      "loss": 0.5258,
+      "step": 32920
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.155106035422095e-06,
+      "loss": 0.4979,
+      "step": 32940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.144626211806038e-06,
+      "loss": 0.4596,
+      "step": 32960
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.13414849810826e-06,
+      "loss": 0.5071,
+      "step": 32980
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.123672906243955e-06,
+      "loss": 0.5356,
+      "step": 33000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.11319944812589e-06,
+      "loss": 0.5452,
+      "step": 33020
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.102728135664411e-06,
+      "loss": 0.5393,
+      "step": 33040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.09225898076742e-06,
+      "loss": 0.5167,
+      "step": 33060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.08179199534037e-06,
+      "loss": 0.4935,
+      "step": 33080
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.071327191286244e-06,
+      "loss": 0.547,
+      "step": 33100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.060864580505542e-06,
+      "loss": 0.5217,
+      "step": 33120
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.050404174896282e-06,
+      "loss": 0.5212,
+      "step": 33140
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.039945986353954e-06,
+      "loss": 0.4672,
+      "step": 33160
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.029490026771545e-06,
+      "loss": 0.4622,
+      "step": 33180
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.019036308039497e-06,
+      "loss": 0.507,
+      "step": 33200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.008584842045706e-06,
+      "loss": 0.5324,
+      "step": 33220
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.998135640675514e-06,
+      "loss": 0.5084,
+      "step": 33240
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.987688715811672e-06,
+      "loss": 0.5583,
+      "step": 33260
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.977244079334357e-06,
+      "loss": 0.5154,
+      "step": 33280
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.966801743121133e-06,
+      "loss": 0.5123,
+      "step": 33300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.95636171904696e-06,
+      "loss": 0.517,
+      "step": 33320
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.945924018984155e-06,
+      "loss": 0.4924,
+      "step": 33340
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.935488654802395e-06,
+      "loss": 0.5687,
+      "step": 33360
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.925055638368713e-06,
+      "loss": 0.5503,
+      "step": 33380
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.914624981547459e-06,
+      "loss": 0.5222,
+      "step": 33400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.90419669620031e-06,
+      "loss": 0.5184,
+      "step": 33420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.893770794186231e-06,
+      "loss": 0.5176,
+      "step": 33440
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.88334728736149e-06,
+      "loss": 0.5423,
+      "step": 33460
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.872926187579625e-06,
+      "loss": 0.5321,
+      "step": 33480
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.862507506691444e-06,
+      "loss": 0.5127,
+      "step": 33500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.852091256545e-06,
+      "loss": 0.4715,
+      "step": 33520
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.841677448985572e-06,
+      "loss": 0.5334,
+      "step": 33540
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.831266095855675e-06,
+      "loss": 0.4881,
+      "step": 33560
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.82085720899503e-06,
+      "loss": 0.5016,
+      "step": 33580
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.810450800240549e-06,
+      "loss": 0.5255,
+      "step": 33600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.800046881426331e-06,
+      "loss": 0.4765,
+      "step": 33620
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.789645464383635e-06,
+      "loss": 0.5301,
+      "step": 33640
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.779246560940887e-06,
+      "loss": 0.4933,
+      "step": 33660
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.768850182923646e-06,
+      "loss": 0.4941,
+      "step": 33680
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.758456342154602e-06,
+      "loss": 0.4816,
+      "step": 33700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.748065050453557e-06,
+      "loss": 0.5509,
+      "step": 33720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.737676319637417e-06,
+      "loss": 0.5269,
+      "step": 33740
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.727290161520176e-06,
+      "loss": 0.4682,
+      "step": 33760
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.716906587912903e-06,
+      "loss": 0.5659,
+      "step": 33780
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.706525610623726e-06,
+      "loss": 0.4829,
+      "step": 33800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.696147241457817e-06,
+      "loss": 0.5116,
+      "step": 33820
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.685771492217387e-06,
+      "loss": 0.4765,
+      "step": 33840
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.675398374701669e-06,
+      "loss": 0.559,
+      "step": 33860
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.665027900706902e-06,
+      "loss": 0.5885,
+      "step": 33880
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.654660082026313e-06,
+      "loss": 0.5742,
+      "step": 33900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.644294930450117e-06,
+      "loss": 0.5308,
+      "step": 33920
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.633932457765491e-06,
+      "loss": 0.5628,
+      "step": 33940
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.623572675756568e-06,
+      "loss": 0.527,
+      "step": 33960
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.613215596204425e-06,
+      "loss": 0.5158,
+      "step": 33980
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.602861230887055e-06,
+      "loss": 0.5089,
+      "step": 34000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.592509591579373e-06,
+      "loss": 0.5192,
+      "step": 34020
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.582160690053194e-06,
+      "loss": 0.536,
+      "step": 34040
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.571814538077216e-06,
+      "loss": 0.5417,
+      "step": 34060
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.5614711474170155e-06,
+      "loss": 0.5583,
+      "step": 34080
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.551130529835016e-06,
+      "loss": 0.5398,
+      "step": 34100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.540792697090502e-06,
+      "loss": 0.5044,
+      "step": 34120
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.530457660939588e-06,
+      "loss": 0.504,
+      "step": 34140
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.5201254331352056e-06,
+      "loss": 0.4996,
+      "step": 34160
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.509796025427088e-06,
+      "loss": 0.534,
+      "step": 34180
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.49946944956177e-06,
+      "loss": 0.5249,
+      "step": 34200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.489145717282559e-06,
+      "loss": 0.4901,
+      "step": 34220
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.478824840329537e-06,
+      "loss": 0.4841,
+      "step": 34240
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.4685068304395366e-06,
+      "loss": 0.4928,
+      "step": 34260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.45819169934612e-06,
+      "loss": 0.551,
+      "step": 34280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.4478794587795855e-06,
+      "loss": 0.507,
+      "step": 34300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.437570120466943e-06,
+      "loss": 0.4897,
+      "step": 34320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.427263696131903e-06,
+      "loss": 0.5578,
+      "step": 34340
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.416960197494856e-06,
+      "loss": 0.5348,
+      "step": 34360
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.406659636272872e-06,
+      "loss": 0.537,
+      "step": 34380
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.396362024179677e-06,
+      "loss": 0.4799,
+      "step": 34400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.386067372925647e-06,
+      "loss": 0.487,
+      "step": 34420
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.375775694217788e-06,
+      "loss": 0.5118,
+      "step": 34440
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.365486999759723e-06,
+      "loss": 0.4961,
+      "step": 34460
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.355201301251689e-06,
+      "loss": 0.5203,
+      "step": 34480
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.344918610390506e-06,
+      "loss": 0.5447,
+      "step": 34500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.334638938869582e-06,
+      "loss": 0.5421,
+      "step": 34520
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.324362298378893e-06,
+      "loss": 0.5051,
+      "step": 34540
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.314088700604958e-06,
+      "loss": 0.4877,
+      "step": 34560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.303818157230841e-06,
+      "loss": 0.4842,
+      "step": 34580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.293550679936135e-06,
+      "loss": 0.532,
+      "step": 34600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.283286280396948e-06,
+      "loss": 0.5177,
+      "step": 34620
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.27302497028588e-06,
+      "loss": 0.537,
+      "step": 34640
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.262766761272022e-06,
+      "loss": 0.5476,
+      "step": 34660
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.252511665020939e-06,
+      "loss": 0.5326,
+      "step": 34680
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.242259693194658e-06,
+      "loss": 0.5295,
+      "step": 34700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.232010857451653e-06,
+      "loss": 0.6052,
+      "step": 34720
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.221765169446819e-06,
+      "loss": 0.5554,
+      "step": 34740
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.211522640831492e-06,
+      "loss": 0.4923,
+      "step": 34760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.2012832832534e-06,
+      "loss": 0.5459,
+      "step": 34780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.1910471083566725e-06,
+      "loss": 0.5057,
+      "step": 34800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.180814127781819e-06,
+      "loss": 0.4881,
+      "step": 34820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.170584353165708e-06,
+      "loss": 0.5223,
+      "step": 34840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.1603577961415704e-06,
+      "loss": 0.5601,
+      "step": 34860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.15013446833898e-06,
+      "loss": 0.5344,
+      "step": 34880
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.139914381383837e-06,
+      "loss": 0.5106,
+      "step": 34900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.129697546898344e-06,
+      "loss": 0.5414,
+      "step": 34920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.119483976501021e-06,
+      "loss": 0.5041,
+      "step": 34940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.10927368180667e-06,
+      "loss": 0.5035,
+      "step": 34960
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.0990666744263646e-06,
+      "loss": 0.5097,
+      "step": 34980
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.0888629659674525e-06,
+      "loss": 0.5195,
+      "step": 35000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.07866256803351e-06,
+      "loss": 0.5062,
+      "step": 35020
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.068465492224361e-06,
+      "loss": 0.514,
+      "step": 35040
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.058271750136054e-06,
+      "loss": 0.5295,
+      "step": 35060
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.04808135336084e-06,
+      "loss": 0.5194,
+      "step": 35080
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.03789431348717e-06,
+      "loss": 0.5081,
+      "step": 35100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.027710642099672e-06,
+      "loss": 0.5042,
+      "step": 35120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.017530350779148e-06,
+      "loss": 0.5264,
+      "step": 35140
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.007353451102557e-06,
+      "loss": 0.4971,
+      "step": 35160
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.997179954642998e-06,
+      "loss": 0.5111,
+      "step": 35180
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.9870098729697015e-06,
+      "loss": 0.5388,
+      "step": 35200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.976843217648013e-06,
+      "loss": 0.5412,
+      "step": 35220
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.966680000239384e-06,
+      "loss": 0.547,
+      "step": 35240
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.956520232301355e-06,
+      "loss": 0.5167,
+      "step": 35260
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.946363925387546e-06,
+      "loss": 0.506,
+      "step": 35280
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.936211091047636e-06,
+      "loss": 0.5407,
+      "step": 35300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.926061740827359e-06,
+      "loss": 0.5601,
+      "step": 35320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.915915886268486e-06,
+      "loss": 0.4812,
+      "step": 35340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.905773538908814e-06,
+      "loss": 0.5248,
+      "step": 35360
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.895634710282153e-06,
+      "loss": 0.51,
+      "step": 35380
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.885499411918304e-06,
+      "loss": 0.5256,
+      "step": 35400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.8753676553430606e-06,
+      "loss": 0.5107,
+      "step": 35420
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.8652394520781865e-06,
+      "loss": 0.5276,
+      "step": 35440
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.855114813641408e-06,
+      "loss": 0.5133,
+      "step": 35460
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.844993751546388e-06,
+      "loss": 0.5299,
+      "step": 35480
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.8348762773027285e-06,
+      "loss": 0.5327,
+      "step": 35500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.824762402415958e-06,
+      "loss": 0.5263,
+      "step": 35520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.814652138387501e-06,
+      "loss": 0.5323,
+      "step": 35540
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.8045454967146854e-06,
+      "loss": 0.5296,
+      "step": 35560
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.794442488890705e-06,
+      "loss": 0.5498,
+      "step": 35580
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.784343126404636e-06,
+      "loss": 0.4546,
+      "step": 35600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.774247420741401e-06,
+      "loss": 0.5471,
+      "step": 35620
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.764155383381772e-06,
+      "loss": 0.5375,
+      "step": 35640
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.754067025802346e-06,
+      "loss": 0.5628,
+      "step": 35660
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.743982359475525e-06,
+      "loss": 0.5029,
+      "step": 35680
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.733901395869527e-06,
+      "loss": 0.4811,
+      "step": 35700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.723824146448352e-06,
+      "loss": 0.5296,
+      "step": 35720
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.713750622671784e-06,
+      "loss": 0.5108,
+      "step": 35740
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.7036808359953585e-06,
+      "loss": 0.5266,
+      "step": 35760
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.693614797870371e-06,
+      "loss": 0.5075,
+      "step": 35780
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.68355251974385e-06,
+      "loss": 0.5385,
+      "step": 35800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.673494013058548e-06,
+      "loss": 0.5294,
+      "step": 35820
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.663439289252932e-06,
+      "loss": 0.5048,
+      "step": 35840
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.653388359761161e-06,
+      "loss": 0.482,
+      "step": 35860
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.6433412360130855e-06,
+      "loss": 0.4721,
+      "step": 35880
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.633297929434224e-06,
+      "loss": 0.5274,
+      "step": 35900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.623258451445758e-06,
+      "loss": 0.5133,
+      "step": 35920
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.613222813464508e-06,
+      "loss": 0.534,
+      "step": 35940
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.603191026902936e-06,
+      "loss": 0.496,
+      "step": 35960
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.59316310316912e-06,
+      "loss": 0.5472,
+      "step": 35980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.583139053666745e-06,
+      "loss": 0.4993,
+      "step": 36000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.573118889795093e-06,
+      "loss": 0.5548,
+      "step": 36020
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.563102622949021e-06,
+      "loss": 0.509,
+      "step": 36040
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.553090264518962e-06,
+      "loss": 0.5209,
+      "step": 36060
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.543081825890901e-06,
+      "loss": 0.5229,
+      "step": 36080
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.533077318446365e-06,
+      "loss": 0.5561,
+      "step": 36100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.523076753562411e-06,
+      "loss": 0.4693,
+      "step": 36120
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.513080142611612e-06,
+      "loss": 0.4867,
+      "step": 36140
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.503087496962044e-06,
+      "loss": 0.5337,
+      "step": 36160
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.493098827977278e-06,
+      "loss": 0.5121,
+      "step": 36180
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.483114147016358e-06,
+      "loss": 0.5246,
+      "step": 36200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.473133465433793e-06,
+      "loss": 0.5049,
+      "step": 36220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.463156794579543e-06,
+      "loss": 0.5014,
+      "step": 36240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.453184145799015e-06,
+      "loss": 0.4795,
+      "step": 36260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.4432155304330335e-06,
+      "loss": 0.5077,
+      "step": 36280
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.433250959817845e-06,
+      "loss": 0.5471,
+      "step": 36300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.42329044528508e-06,
+      "loss": 0.5006,
+      "step": 36320
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.413333998161772e-06,
+      "loss": 0.518,
+      "step": 36340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.4033816297703245e-06,
+      "loss": 0.5612,
+      "step": 36360
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.393433351428502e-06,
+      "loss": 0.4518,
+      "step": 36380
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.383489174449422e-06,
+      "loss": 0.5344,
+      "step": 36400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.373549110141528e-06,
+      "loss": 0.5152,
+      "step": 36420
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.363613169808593e-06,
+      "loss": 0.4774,
+      "step": 36440
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.353681364749703e-06,
+      "loss": 0.481,
+      "step": 36460
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.343753706259239e-06,
+      "loss": 0.5457,
+      "step": 36480
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.333830205626861e-06,
+      "loss": 0.5329,
+      "step": 36500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.323910874137509e-06,
+      "loss": 0.4489,
+      "step": 36520
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.313995723071377e-06,
+      "loss": 0.4993,
+      "step": 36540
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.3040847637039075e-06,
+      "loss": 0.5238,
+      "step": 36560
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.294178007305776e-06,
+      "loss": 0.507,
+      "step": 36580
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.284275465142875e-06,
+      "loss": 0.5401,
+      "step": 36600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.27437714847631e-06,
+      "loss": 0.4962,
+      "step": 36620
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.2644830685623756e-06,
+      "loss": 0.4884,
+      "step": 36640
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.254593236652554e-06,
+      "loss": 0.5003,
+      "step": 36660
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.244707663993494e-06,
+      "loss": 0.4925,
+      "step": 36680
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.2348263618269975e-06,
+      "loss": 0.4712,
+      "step": 36700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.224949341390017e-06,
+      "loss": 0.5206,
+      "step": 36720
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.215076613914628e-06,
+      "loss": 0.4793,
+      "step": 36740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.205208190628034e-06,
+      "loss": 0.5046,
+      "step": 36760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.1953440827525306e-06,
+      "loss": 0.5188,
+      "step": 36780
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.185484301505518e-06,
+      "loss": 0.5281,
+      "step": 36800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.175628858099471e-06,
+      "loss": 0.4982,
+      "step": 36820
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.165777763741932e-06,
+      "loss": 0.5057,
+      "step": 36840
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.1559310296355005e-06,
+      "loss": 0.5274,
+      "step": 36860
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.146088666977804e-06,
+      "loss": 0.487,
+      "step": 36880
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.1362506869615205e-06,
+      "loss": 0.5438,
+      "step": 36900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.126417100774329e-06,
+      "loss": 0.5017,
+      "step": 36920
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.116587919598915e-06,
+      "loss": 0.5062,
+      "step": 36940
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.106763154612963e-06,
+      "loss": 0.5229,
+      "step": 36960
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.096942816989116e-06,
+      "loss": 0.4945,
+      "step": 36980
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.0871269178949986e-06,
+      "loss": 0.569,
+      "step": 37000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.0773154684931875e-06,
+      "loss": 0.5008,
+      "step": 37020
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.0675084799411975e-06,
+      "loss": 0.5091,
+      "step": 37040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.057705963391461e-06,
+      "loss": 0.5372,
+      "step": 37060
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.047907929991333e-06,
+      "loss": 0.5228,
+      "step": 37080
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.038114390883073e-06,
+      "loss": 0.5057,
+      "step": 37100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.028325357203822e-06,
+      "loss": 0.4866,
+      "step": 37120
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.018540840085611e-06,
+      "loss": 0.5037,
+      "step": 37140
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.008760850655316e-06,
+      "loss": 0.5303,
+      "step": 37160
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.998985400034676e-06,
+      "loss": 0.507,
+      "step": 37180
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.989214499340267e-06,
+      "loss": 0.4896,
+      "step": 37200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.979448159683491e-06,
+      "loss": 0.5309,
+      "step": 37220
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.96968639217056e-06,
+      "loss": 0.4934,
+      "step": 37240
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.959929207902488e-06,
+      "loss": 0.4939,
+      "step": 37260
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.950176617975079e-06,
+      "loss": 0.5182,
+      "step": 37280
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.9404286334789115e-06,
+      "loss": 0.5739,
+      "step": 37300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.930685265499329e-06,
+      "loss": 0.4918,
+      "step": 37320
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.9209465251164175e-06,
+      "loss": 0.5235,
+      "step": 37340
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.911212423405008e-06,
+      "loss": 0.5176,
+      "step": 37360
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.901482971434652e-06,
+      "loss": 0.5081,
+      "step": 37380
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.891758180269618e-06,
+      "loss": 0.4788,
+      "step": 37400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.882038060968871e-06,
+      "loss": 0.528,
+      "step": 37420
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.872322624586061e-06,
+      "loss": 0.5005,
+      "step": 37440
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.862611882169518e-06,
+      "loss": 0.4987,
+      "step": 37460
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.852905844762229e-06,
+      "loss": 0.4889,
+      "step": 37480
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.8432045234018365e-06,
+      "loss": 0.4652,
+      "step": 37500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.833507929120612e-06,
+      "loss": 0.5429,
+      "step": 37520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.823816072945458e-06,
+      "loss": 0.5374,
+      "step": 37540
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.814128965897887e-06,
+      "loss": 0.4909,
+      "step": 37560
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.804446618994009e-06,
+      "loss": 0.498,
+      "step": 37580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.794769043244527e-06,
+      "loss": 0.4974,
+      "step": 37600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.785096249654705e-06,
+      "loss": 0.4769,
+      "step": 37620
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.775428249224382e-06,
+      "loss": 0.5101,
+      "step": 37640
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.765765052947941e-06,
+      "loss": 0.4938,
+      "step": 37660
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.756106671814301e-06,
+      "loss": 0.4921,
+      "step": 37680
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.7464531168069096e-06,
+      "loss": 0.5029,
+      "step": 37700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.7368043989037206e-06,
+      "loss": 0.5164,
+      "step": 37720
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.7271605290771895e-06,
+      "loss": 0.5038,
+      "step": 37740
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.717521518294259e-06,
+      "loss": 0.5467,
+      "step": 37760
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.70788737751635e-06,
+      "loss": 0.4912,
+      "step": 37780
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.698258117699334e-06,
+      "loss": 0.5349,
+      "step": 37800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.68863374979354e-06,
+      "loss": 0.5092,
+      "step": 37820
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.6790142847437355e-06,
+      "loss": 0.5157,
+      "step": 37840
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.669399733489108e-06,
+      "loss": 0.5729,
+      "step": 37860
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.659790106963261e-06,
+      "loss": 0.5483,
+      "step": 37880
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.650185416094194e-06,
+      "loss": 0.5188,
+      "step": 37900
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.640585671804296e-06,
+      "loss": 0.5054,
+      "step": 37920
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.63099088501033e-06,
+      "loss": 0.5646,
+      "step": 37940
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.621401066623421e-06,
+      "loss": 0.5179,
+      "step": 37960
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.611816227549048e-06,
+      "loss": 0.4797,
+      "step": 37980
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.602236378687018e-06,
+      "loss": 0.5401,
+      "step": 38000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.59266153093147e-06,
+      "loss": 0.4801,
+      "step": 38020
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.583091695170857e-06,
+      "loss": 0.4999,
+      "step": 38040
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.5735268822879344e-06,
+      "loss": 0.4958,
+      "step": 38060
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.563967103159733e-06,
+      "loss": 0.4913,
+      "step": 38080
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.554412368657572e-06,
+      "loss": 0.5331,
+      "step": 38100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.544862689647024e-06,
+      "loss": 0.5276,
+      "step": 38120
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.535318076987928e-06,
+      "loss": 0.5382,
+      "step": 38140
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.52577854153435e-06,
+      "loss": 0.508,
+      "step": 38160
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.516244094134578e-06,
+      "loss": 0.5848,
+      "step": 38180
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.506714745631124e-06,
+      "loss": 0.5551,
+      "step": 38200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.497190506860695e-06,
+      "loss": 0.5111,
+      "step": 38220
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.487671388654192e-06,
+      "loss": 0.5263,
+      "step": 38240
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.478157401836694e-06,
+      "loss": 0.5426,
+      "step": 38260
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.468648557227434e-06,
+      "loss": 0.5187,
+      "step": 38280
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.459144865639806e-06,
+      "loss": 0.5232,
+      "step": 38300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.449646337881345e-06,
+      "loss": 0.5209,
+      "step": 38320
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.440152984753709e-06,
+      "loss": 0.4915,
+      "step": 38340
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.430664817052678e-06,
+      "loss": 0.5327,
+      "step": 38360
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.421181845568125e-06,
+      "loss": 0.4883,
+      "step": 38380
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.411704081084025e-06,
+      "loss": 0.5274,
+      "step": 38400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.402231534378424e-06,
+      "loss": 0.5467,
+      "step": 38420
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.392764216223441e-06,
+      "loss": 0.5268,
+      "step": 38440
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.383302137385238e-06,
+      "loss": 0.5262,
+      "step": 38460
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.373845308624028e-06,
+      "loss": 0.541,
+      "step": 38480
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.364393740694054e-06,
+      "loss": 0.5145,
+      "step": 38500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.3549474443435726e-06,
+      "loss": 0.4997,
+      "step": 38520
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.345506430314845e-06,
+      "loss": 0.4927,
+      "step": 38540
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.33607070934413e-06,
+      "loss": 0.516,
+      "step": 38560
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.326640292161665e-06,
+      "loss": 0.549,
+      "step": 38580
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.317215189491651e-06,
+      "loss": 0.5273,
+      "step": 38600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.307795412052259e-06,
+      "loss": 0.5299,
+      "step": 38620
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.298380970555584e-06,
+      "loss": 0.5154,
+      "step": 38640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.288971875707669e-06,
+      "loss": 0.5577,
+      "step": 38660
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.279568138208473e-06,
+      "loss": 0.497,
+      "step": 38680
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.270169768751858e-06,
+      "loss": 0.4864,
+      "step": 38700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.260776778025591e-06,
+      "loss": 0.494,
+      "step": 38720
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.251389176711307e-06,
+      "loss": 0.4946,
+      "step": 38740
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.242006975484528e-06,
+      "loss": 0.4608,
+      "step": 38760
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.2326301850146285e-06,
+      "loss": 0.5286,
+      "step": 38780
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.223258815964833e-06,
+      "loss": 0.511,
+      "step": 38800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.213892878992191e-06,
+      "loss": 0.5245,
+      "step": 38820
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.204532384747584e-06,
+      "loss": 0.5,
+      "step": 38840
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.195177343875701e-06,
+      "loss": 0.5208,
+      "step": 38860
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.185827767015031e-06,
+      "loss": 0.5128,
+      "step": 38880
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.176483664797852e-06,
+      "loss": 0.5312,
+      "step": 38900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.1671450478502064e-06,
+      "loss": 0.4755,
+      "step": 38920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.157811926791905e-06,
+      "loss": 0.4955,
+      "step": 38940
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.1484843122365125e-06,
+      "loss": 0.4997,
+      "step": 38960
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.139162214791322e-06,
+      "loss": 0.5367,
+      "step": 38980
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.129845645057372e-06,
+      "loss": 0.5051,
+      "step": 39000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.120534613629389e-06,
+      "loss": 0.5145,
+      "step": 39020
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.1112291310958166e-06,
+      "loss": 0.4681,
+      "step": 39040
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.101929208038786e-06,
+      "loss": 0.5096,
+      "step": 39060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.09263485503411e-06,
+      "loss": 0.4984,
+      "step": 39080
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.083346082651255e-06,
+      "loss": 0.4949,
+      "step": 39100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.074062901453352e-06,
+      "loss": 0.4977,
+      "step": 39120
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.06478532199717e-06,
+      "loss": 0.5075,
+      "step": 39140
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.05551335483311e-06,
+      "loss": 0.4985,
+      "step": 39160
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.04624701050519e-06,
+      "loss": 0.4894,
+      "step": 39180
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.03698629955103e-06,
+      "loss": 0.4938,
+      "step": 39200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.027731232501849e-06,
+      "loss": 0.5228,
+      "step": 39220
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.018481819882446e-06,
+      "loss": 0.4612,
+      "step": 39240
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.009238072211187e-06,
+      "loss": 0.5362,
+      "step": 39260
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.4859,
+      "step": 39280
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.990767613754358e-06,
+      "loss": 0.49,
+      "step": 39300
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.981540923973263e-06,
+      "loss": 0.5071,
+      "step": 39320
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.972319941149245e-06,
+      "loss": 0.5129,
+      "step": 39340
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.963104675768345e-06,
+      "loss": 0.5516,
+      "step": 39360
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.953895138310091e-06,
+      "loss": 0.4968,
+      "step": 39380
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.944691339247515e-06,
+      "loss": 0.5478,
+      "step": 39400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.9354932890471096e-06,
+      "loss": 0.5459,
+      "step": 39420
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.926300998168835e-06,
+      "loss": 0.518,
+      "step": 39440
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.917114477066104e-06,
+      "loss": 0.4936,
+      "step": 39460
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.907933736185757e-06,
+      "loss": 0.4986,
+      "step": 39480
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.898758785968075e-06,
+      "loss": 0.4951,
+      "step": 39500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.889589636846747e-06,
+      "loss": 0.5219,
+      "step": 39520
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.880426299248867e-06,
+      "loss": 0.4831,
+      "step": 39540
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.871268783594922e-06,
+      "loss": 0.503,
+      "step": 39560
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.86211710029877e-06,
+      "loss": 0.5001,
+      "step": 39580
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.852971259767642e-06,
+      "loss": 0.4996,
+      "step": 39600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.843831272402126e-06,
+      "loss": 0.469,
+      "step": 39620
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.83469714859616e-06,
+      "loss": 0.5212,
+      "step": 39640
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.825568898736999e-06,
+      "loss": 0.4983,
+      "step": 39660
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.816446533205225e-06,
+      "loss": 0.5324,
+      "step": 39680
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.807330062374733e-06,
+      "loss": 0.5249,
+      "step": 39700
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.79821949661271e-06,
+      "loss": 0.4734,
+      "step": 39720
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.789114846279631e-06,
+      "loss": 0.5388,
+      "step": 39740
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.780016121729236e-06,
+      "loss": 0.5085,
+      "step": 39760
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7709233333085345e-06,
+      "loss": 0.5055,
+      "step": 39780
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.761836491357782e-06,
+      "loss": 0.4652,
+      "step": 39800
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7527556062104756e-06,
+      "loss": 0.524,
+      "step": 39820
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7436806881933304e-06,
+      "loss": 0.4764,
+      "step": 39840
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.734611747626285e-06,
+      "loss": 0.4892,
+      "step": 39860
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.7255487948224754e-06,
+      "loss": 0.4889,
+      "step": 39880
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.716491840088227e-06,
+      "loss": 0.533,
+      "step": 39900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.707440893723052e-06,
+      "loss": 0.5227,
+      "step": 39920
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.698395966019615e-06,
+      "loss": 0.5253,
+      "step": 39940
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.689357067263751e-06,
+      "loss": 0.5263,
+      "step": 39960
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.680324207734434e-06,
+      "loss": 0.509,
+      "step": 39980
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.671297397703766e-06,
+      "loss": 0.5364,
+      "step": 40000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.662276647436977e-06,
+      "loss": 0.4903,
+      "step": 40020
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.653261967192398e-06,
+      "loss": 0.5374,
+      "step": 40040
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.644253367221467e-06,
+      "loss": 0.478,
+      "step": 40060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.635250857768696e-06,
+      "loss": 0.4984,
+      "step": 40080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.62625444907168e-06,
+      "loss": 0.4795,
+      "step": 40100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6172641513610765e-06,
+      "loss": 0.4602,
+      "step": 40120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.608279974860581e-06,
+      "loss": 0.4751,
+      "step": 40140
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.599301929786943e-06,
+      "loss": 0.5241,
+      "step": 40160
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.5903300263499315e-06,
+      "loss": 0.504,
+      "step": 40180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.581364274752338e-06,
+      "loss": 0.5306,
+      "step": 40200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.572404685189946e-06,
+      "loss": 0.5155,
+      "step": 40220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.563451267851542e-06,
+      "loss": 0.4783,
+      "step": 40240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.554504032918888e-06,
+      "loss": 0.464,
+      "step": 40260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.545562990566725e-06,
+      "loss": 0.4934,
+      "step": 40280
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.536628150962744e-06,
+      "loss": 0.4472,
+      "step": 40300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.527699524267576e-06,
+      "loss": 0.4941,
+      "step": 40320
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.518777120634798e-06,
+      "loss": 0.4424,
+      "step": 40340
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.509860950210905e-06,
+      "loss": 0.5254,
+      "step": 40360
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.500951023135308e-06,
+      "loss": 0.4916,
+      "step": 40380
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.492047349540308e-06,
+      "loss": 0.5132,
+      "step": 40400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.483149939551105e-06,
+      "loss": 0.5309,
+      "step": 40420
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.474258803285774e-06,
+      "loss": 0.5116,
+      "step": 40440
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.465373950855249e-06,
+      "loss": 0.4641,
+      "step": 40460
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.4564953923633295e-06,
+      "loss": 0.478,
+      "step": 40480
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.4476231379066445e-06,
+      "loss": 0.4497,
+      "step": 40500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.4387571975746655e-06,
+      "loss": 0.5384,
+      "step": 40520
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.429897581449677e-06,
+      "loss": 0.4797,
+      "step": 40540
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.4210442996067724e-06,
+      "loss": 0.5034,
+      "step": 40560
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.412197362113848e-06,
+      "loss": 0.4553,
+      "step": 40580
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.403356779031572e-06,
+      "loss": 0.518,
+      "step": 40600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.394522560413399e-06,
+      "loss": 0.4733,
+      "step": 40620
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.385694716305541e-06,
+      "loss": 0.4931,
+      "step": 40640
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.376873256746962e-06,
+      "loss": 0.5156,
+      "step": 40660
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.368058191769363e-06,
+      "loss": 0.5206,
+      "step": 40680
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.359249531397177e-06,
+      "loss": 0.5079,
+      "step": 40700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.350447285647552e-06,
+      "loss": 0.4872,
+      "step": 40720
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.341651464530341e-06,
+      "loss": 0.4953,
+      "step": 40740
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.332862078048095e-06,
+      "loss": 0.5342,
+      "step": 40760
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.324079136196034e-06,
+      "loss": 0.5427,
+      "step": 40780
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.315302648962066e-06,
+      "loss": 0.5343,
+      "step": 40800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.30653262632675e-06,
+      "loss": 0.5144,
+      "step": 40820
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.297769078263296e-06,
+      "loss": 0.4854,
+      "step": 40840
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2890120147375545e-06,
+      "loss": 0.5271,
+      "step": 40860
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.280261445707985e-06,
+      "loss": 0.5053,
+      "step": 40880
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.271517381125687e-06,
+      "loss": 0.5188,
+      "step": 40900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.262779830934346e-06,
+      "loss": 0.5301,
+      "step": 40920
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.254048805070248e-06,
+      "loss": 0.4891,
+      "step": 40940
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2453243134622465e-06,
+      "loss": 0.4493,
+      "step": 40960
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2366063660317805e-06,
+      "loss": 0.4975,
+      "step": 40980
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2278949726928354e-06,
+      "loss": 0.5603,
+      "step": 41000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.219190143351951e-06,
+      "loss": 0.5002,
+      "step": 41020
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.210491887908201e-06,
+      "loss": 0.5157,
+      "step": 41040
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.201800216253176e-06,
+      "loss": 0.5158,
+      "step": 41060
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.193115138270986e-06,
+      "loss": 0.5076,
+      "step": 41080
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.184436663838245e-06,
+      "loss": 0.4968,
+      "step": 41100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.175764802824049e-06,
+      "loss": 0.5185,
+      "step": 41120
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.167099565089989e-06,
+      "loss": 0.5321,
+      "step": 41140
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.158440960490104e-06,
+      "loss": 0.4719,
+      "step": 41160
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.149788998870902e-06,
+      "loss": 0.525,
+      "step": 41180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.141143690071335e-06,
+      "loss": 0.497,
+      "step": 41200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.132505043922792e-06,
+      "loss": 0.4672,
+      "step": 41220
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.123873070249074e-06,
+      "loss": 0.4633,
+      "step": 41240
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.115247778866407e-06,
+      "loss": 0.4851,
+      "step": 41260
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.106629179583412e-06,
+      "loss": 0.4671,
+      "step": 41280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.0980172822010985e-06,
+      "loss": 0.4747,
+      "step": 41300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.0894120965128615e-06,
+      "loss": 0.5273,
+      "step": 41320
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.080813632304455e-06,
+      "loss": 0.4695,
+      "step": 41340
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.072221899353995e-06,
+      "loss": 0.5268,
+      "step": 41360
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.063636907431942e-06,
+      "loss": 0.5052,
+      "step": 41380
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.055058666301087e-06,
+      "loss": 0.5012,
+      "step": 41400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.046487185716552e-06,
+      "loss": 0.5278,
+      "step": 41420
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.037922475425759e-06,
+      "loss": 0.4867,
+      "step": 41440
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.029364545168438e-06,
+      "loss": 0.4981,
+      "step": 41460
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.020813404676612e-06,
+      "loss": 0.4956,
+      "step": 41480
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.012269063674581e-06,
+      "loss": 0.4961,
+      "step": 41500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.0037315318789e-06,
+      "loss": 0.4949,
+      "step": 41520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.9952008189984025e-06,
+      "loss": 0.4913,
+      "step": 41540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.986676934734151e-06,
+      "loss": 0.554,
+      "step": 41560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.978159888779451e-06,
+      "loss": 0.4984,
+      "step": 41580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.96964969081983e-06,
+      "loss": 0.5367,
+      "step": 41600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.961146350533021e-06,
+      "loss": 0.4877,
+      "step": 41620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.9526498775889645e-06,
+      "loss": 0.5043,
+      "step": 41640
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.944160281649793e-06,
+      "loss": 0.4978,
+      "step": 41660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.935677572369817e-06,
+      "loss": 0.5095,
+      "step": 41680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.927201759395518e-06,
+      "loss": 0.4937,
+      "step": 41700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.918732852365524e-06,
+      "loss": 0.5151,
+      "step": 41720
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.910270860910623e-06,
+      "loss": 0.4712,
+      "step": 41740
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.901815794653729e-06,
+      "loss": 0.4514,
+      "step": 41760
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.893367663209896e-06,
+      "loss": 0.5102,
+      "step": 41780
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.884926476186269e-06,
+      "loss": 0.4833,
+      "step": 41800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.876492243182113e-06,
+      "loss": 0.4814,
+      "step": 41820
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.8680649737887784e-06,
+      "loss": 0.5291,
+      "step": 41840
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.859644677589699e-06,
+      "loss": 0.5029,
+      "step": 41860
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.851231364160379e-06,
+      "loss": 0.4675,
+      "step": 41880
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.842825043068377e-06,
+      "loss": 0.5084,
+      "step": 41900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.834425723873303e-06,
+      "loss": 0.4999,
+      "step": 41920
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.826033416126807e-06,
+      "loss": 0.4788,
+      "step": 41940
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.817648129372563e-06,
+      "loss": 0.5149,
+      "step": 41960
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.809269873146263e-06,
+      "loss": 0.5007,
+      "step": 41980
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.800898656975599e-06,
+      "loss": 0.477,
+      "step": 42000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.792534490380263e-06,
+      "loss": 0.4822,
+      "step": 42020
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.7841773828719277e-06,
+      "loss": 0.4666,
+      "step": 42040
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.775827343954239e-06,
+      "loss": 0.4678,
+      "step": 42060
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.767484383122799e-06,
+      "loss": 0.4648,
+      "step": 42080
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.759148509865168e-06,
+      "loss": 0.4941,
+      "step": 42100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.750819733660844e-06,
+      "loss": 0.5476,
+      "step": 42120
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.7424980639812534e-06,
+      "loss": 0.4731,
+      "step": 42140
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.734183510289742e-06,
+      "loss": 0.5135,
+      "step": 42160
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.725876082041563e-06,
+      "loss": 0.498,
+      "step": 42180
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.717575788683866e-06,
+      "loss": 0.4935,
+      "step": 42200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.7092826396556882e-06,
+      "loss": 0.5092,
+      "step": 42220
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.700996644387944e-06,
+      "loss": 0.4793,
+      "step": 42240
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.6927178123034034e-06,
+      "loss": 0.4633,
+      "step": 42260
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.684446152816702e-06,
+      "loss": 0.5189,
+      "step": 42280
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.6761816753343114e-06,
+      "loss": 0.4972,
+      "step": 42300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.6679243892545414e-06,
+      "loss": 0.5158,
+      "step": 42320
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.659674303967522e-06,
+      "loss": 0.5912,
+      "step": 42340
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.6514314288551877e-06,
+      "loss": 0.4993,
+      "step": 42360
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.6431957732912793e-06,
+      "loss": 0.4731,
+      "step": 42380
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.634967346641334e-06,
+      "loss": 0.5195,
+      "step": 42400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.6267461582626584e-06,
+      "loss": 0.5218,
+      "step": 42420
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.6185322175043336e-06,
+      "loss": 0.4498,
+      "step": 42440
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.6103255337071918e-06,
+      "loss": 0.5465,
+      "step": 42460
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.602126116203819e-06,
+      "loss": 0.5014,
+      "step": 42480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5939339743185376e-06,
+      "loss": 0.5211,
+      "step": 42500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5857491173673976e-06,
+      "loss": 0.4967,
+      "step": 42520
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5775715546581546e-06,
+      "loss": 0.5686,
+      "step": 42540
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5694012954902803e-06,
+      "loss": 0.4815,
+      "step": 42560
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5612383491549384e-06,
+      "loss": 0.4797,
+      "step": 42580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.553082724934973e-06,
+      "loss": 0.4874,
+      "step": 42600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.544934432104905e-06,
+      "loss": 0.4657,
+      "step": 42620
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5367934799309167e-06,
+      "loss": 0.5194,
+      "step": 42640
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5286598776708436e-06,
+      "loss": 0.4837,
+      "step": 42660
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5205336345741603e-06,
+      "loss": 0.4811,
+      "step": 42680
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.512414759881976e-06,
+      "loss": 0.4934,
+      "step": 42700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5043032628270225e-06,
+      "loss": 0.4978,
+      "step": 42720
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.49619915263363e-06,
+      "loss": 0.4769,
+      "step": 42740
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.488102438517742e-06,
+      "loss": 0.5403,
+      "step": 42760
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4800131296868842e-06,
+      "loss": 0.5017,
+      "step": 42780
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4719312353401647e-06,
+      "loss": 0.4866,
+      "step": 42800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4638567646682576e-06,
+      "loss": 0.4903,
+      "step": 42820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4557897268533935e-06,
+      "loss": 0.4781,
+      "step": 42840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.4477301310693544e-06,
+      "loss": 0.4909,
+      "step": 42860
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.4396779864814568e-06,
+      "loss": 0.5015,
+      "step": 42880
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.4316333022465465e-06,
+      "loss": 0.4896,
+      "step": 42900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.423596087512976e-06,
+      "loss": 0.5027,
+      "step": 42920
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.415566351420615e-06,
+      "loss": 0.497,
+      "step": 42940
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.407544103100824e-06,
+      "loss": 0.4664,
+      "step": 42960
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.399529351676447e-06,
+      "loss": 0.4745,
+      "step": 42980
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.39152210626181e-06,
+      "loss": 0.5351,
+      "step": 43000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3835223759626845e-06,
+      "loss": 0.4961,
+      "step": 43020
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3755301698763197e-06,
+      "loss": 0.5245,
+      "step": 43040
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.367545497091396e-06,
+      "loss": 0.5402,
+      "step": 43060
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3595683666880286e-06,
+      "loss": 0.5098,
+      "step": 43080
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3515987877377498e-06,
+      "loss": 0.4786,
+      "step": 43100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.343636769303512e-06,
+      "loss": 0.4555,
+      "step": 43120
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.3356823204396706e-06,
+      "loss": 0.4998,
+      "step": 43140
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3277354501919677e-06,
+      "loss": 0.4714,
+      "step": 43160
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.319796167597533e-06,
+      "loss": 0.4957,
+      "step": 43180
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.3118644816848577e-06,
+      "loss": 0.4871,
+      "step": 43200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.303940401473804e-06,
+      "loss": 0.4978,
+      "step": 43220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2960239359755808e-06,
+      "loss": 0.5106,
+      "step": 43240
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.288115094192734e-06,
+      "loss": 0.4918,
+      "step": 43260
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2802138851191558e-06,
+      "loss": 0.5428,
+      "step": 43280
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2723203177400352e-06,
+      "loss": 0.5042,
+      "step": 43300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.264434401031887e-06,
+      "loss": 0.4565,
+      "step": 43320
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2565561439625205e-06,
+      "loss": 0.5296,
+      "step": 43340
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.248685555491039e-06,
+      "loss": 0.4808,
+      "step": 43360
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.240822644567815e-06,
+      "loss": 0.5009,
+      "step": 43380
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2329674201344997e-06,
+      "loss": 0.522,
+      "step": 43400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.2251198911240002e-06,
+      "loss": 0.5334,
+      "step": 43420
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.217280066460472e-06,
+      "loss": 0.4565,
+      "step": 43440
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2094479550593107e-06,
+      "loss": 0.5016,
+      "step": 43460
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.201623565827139e-06,
+      "loss": 0.4863,
+      "step": 43480
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1938069076617983e-06,
+      "loss": 0.4797,
+      "step": 43500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1859979894523375e-06,
+      "loss": 0.5096,
+      "step": 43520
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1781968200790094e-06,
+      "loss": 0.5187,
+      "step": 43540
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.170403408413243e-06,
+      "loss": 0.5091,
+      "step": 43560
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1626177633176557e-06,
+      "loss": 0.5026,
+      "step": 43580
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.154839893646028e-06,
+      "loss": 0.5275,
+      "step": 43600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1470698082433014e-06,
+      "loss": 0.4637,
+      "step": 43620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.139307515945563e-06,
+      "loss": 0.4792,
+      "step": 43640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1315530255800384e-06,
+      "loss": 0.5096,
+      "step": 43660
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1238063459650802e-06,
+      "loss": 0.5076,
+      "step": 43680
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.116067485910159e-06,
+      "loss": 0.4875,
+      "step": 43700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1083364542158535e-06,
+      "loss": 0.5021,
+      "step": 43720
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1006132596738426e-06,
+      "loss": 0.5041,
+      "step": 43740
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.092897911066883e-06,
+      "loss": 0.4867,
+      "step": 43760
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.0851904171688196e-06,
+      "loss": 0.5268,
+      "step": 43780
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.0774907867445623e-06,
+      "loss": 0.5006,
+      "step": 43800
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.0697990285500802e-06,
+      "loss": 0.4879,
+      "step": 43820
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.06211515133238e-06,
+      "loss": 0.5086,
+      "step": 43840
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.054439163829519e-06,
+      "loss": 0.4727,
+      "step": 43860
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.0467710747705735e-06,
+      "loss": 0.4926,
+      "step": 43880
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.0391108928756486e-06,
+      "loss": 0.508,
+      "step": 43900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.0314586268558486e-06,
+      "loss": 0.476,
+      "step": 43920
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.0238142854132722e-06,
+      "loss": 0.5425,
+      "step": 43940
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.0161778772410153e-06,
+      "loss": 0.4988,
+      "step": 43960
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.0085494110231493e-06,
+      "loss": 0.4909,
+      "step": 43980
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.0009288954347115e-06,
+      "loss": 0.5108,
+      "step": 44000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.9933163391417042e-06,
+      "loss": 0.4601,
+      "step": 44020
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.985711750801068e-06,
+      "loss": 0.5138,
+      "step": 44040
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.97811513906069e-06,
+      "loss": 0.4738,
+      "step": 44060
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.9705265125593863e-06,
+      "loss": 0.5138,
+      "step": 44080
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.9629458799268897e-06,
+      "loss": 0.4846,
+      "step": 44100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.955373249783844e-06,
+      "loss": 0.484,
+      "step": 44120
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.947808630741791e-06,
+      "loss": 0.5328,
+      "step": 44140
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.9402520314031646e-06,
+      "loss": 0.4963,
+      "step": 44160
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.932703460361277e-06,
+      "loss": 0.434,
+      "step": 44180
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.9251629262003135e-06,
+      "loss": 0.4835,
+      "step": 44200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.9176304374953123e-06,
+      "loss": 0.5047,
+      "step": 44220
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.910106002812172e-06,
+      "loss": 0.4763,
+      "step": 44240
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.902589630707624e-06,
+      "loss": 0.4903,
+      "step": 44260
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.895081329729239e-06,
+      "loss": 0.4682,
+      "step": 44280
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.8875811084154025e-06,
+      "loss": 0.4581,
+      "step": 44300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.8800889752953156e-06,
+      "loss": 0.4711,
+      "step": 44320
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.872604938888981e-06,
+      "loss": 0.5166,
+      "step": 44340
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.8651290077071914e-06,
+      "loss": 0.5098,
+      "step": 44360
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.8576611902515283e-06,
+      "loss": 0.4655,
+      "step": 44380
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.8502014950143376e-06,
+      "loss": 0.4898,
+      "step": 44400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.842749930478734e-06,
+      "loss": 0.5059,
+      "step": 44420
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.8353065051185867e-06,
+      "loss": 0.4923,
+      "step": 44440
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.827871227398508e-06,
+      "loss": 0.5248,
+      "step": 44460
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.820444105773845e-06,
+      "loss": 0.4879,
+      "step": 44480
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.8130251486906666e-06,
+      "loss": 0.4507,
+      "step": 44500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.805614364585758e-06,
+      "loss": 0.5031,
+      "step": 44520
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.7982117618866177e-06,
+      "loss": 0.4808,
+      "step": 44540
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.790817349011432e-06,
+      "loss": 0.4506,
+      "step": 44560
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.783431134369079e-06,
+      "loss": 0.5022,
+      "step": 44580
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.776053126359105e-06,
+      "loss": 0.5251,
+      "step": 44600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.7686833333717334e-06,
+      "loss": 0.4724,
+      "step": 44620
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.761321763787841e-06,
+      "loss": 0.4463,
+      "step": 44640
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.753968425978958e-06,
+      "loss": 0.457,
+      "step": 44660
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.746623328307243e-06,
+      "loss": 0.4865,
+      "step": 44680
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.739286479125494e-06,
+      "loss": 0.5148,
+      "step": 44700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7319578867771237e-06,
+      "loss": 0.493,
+      "step": 44720
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7246375595961584e-06,
+      "loss": 0.4611,
+      "step": 44740
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7173255059072234e-06,
+      "loss": 0.4909,
+      "step": 44760
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7100217340255352e-06,
+      "loss": 0.5355,
+      "step": 44780
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7027262522568944e-06,
+      "loss": 0.4923,
+      "step": 44800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.6954390688976706e-06,
+      "loss": 0.5313,
+      "step": 44820
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.6881601922348e-06,
+      "loss": 0.4939,
+      "step": 44840
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.6808896305457733e-06,
+      "loss": 0.529,
+      "step": 44860
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.6736273920986166e-06,
+      "loss": 0.4512,
+      "step": 44880
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.6663734851519017e-06,
+      "loss": 0.4795,
+      "step": 44900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.65912791795472e-06,
+      "loss": 0.5323,
+      "step": 44920
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.651890698746682e-06,
+      "loss": 0.4636,
+      "step": 44940
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6446618357579035e-06,
+      "loss": 0.4466,
+      "step": 44960
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6374413372089968e-06,
+      "loss": 0.5066,
+      "step": 44980
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.630229211311064e-06,
+      "loss": 0.4642,
+      "step": 45000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.623025466265684e-06,
+      "loss": 0.4945,
+      "step": 45020
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6158301102649085e-06,
+      "loss": 0.4825,
+      "step": 45040
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6086431514912438e-06,
+      "loss": 0.4915,
+      "step": 45060
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6014645981176488e-06,
+      "loss": 0.4504,
+      "step": 45080
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.594294458307528e-06,
+      "loss": 0.4665,
+      "step": 45100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5871327402147173e-06,
+      "loss": 0.5008,
+      "step": 45120
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5799794519834632e-06,
+      "loss": 0.4345,
+      "step": 45140
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5728346017484453e-06,
+      "loss": 0.4824,
+      "step": 45160
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.565698197634735e-06,
+      "loss": 0.5238,
+      "step": 45180
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5585702477578024e-06,
+      "loss": 0.4494,
+      "step": 45200
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5514507602235063e-06,
+      "loss": 0.4856,
+      "step": 45220
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5443397431280703e-06,
+      "loss": 0.4827,
+      "step": 45240
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5372372045581006e-06,
+      "loss": 0.4832,
+      "step": 45260
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.530143152590553e-06,
+      "loss": 0.4766,
+      "step": 45280
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.523057595292734e-06,
+      "loss": 0.544,
+      "step": 45300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.515980540722295e-06,
+      "loss": 0.4732,
+      "step": 45320
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.508911996927207e-06,
+      "loss": 0.478,
+      "step": 45340
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5018519719457725e-06,
+      "loss": 0.5033,
+      "step": 45360
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4948004738066e-06,
+      "loss": 0.5514,
+      "step": 45380
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4877575105286146e-06,
+      "loss": 0.515,
+      "step": 45400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4807230901210155e-06,
+      "loss": 0.4947,
+      "step": 45420
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4736972205832997e-06,
+      "loss": 0.4861,
+      "step": 45440
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4666799099052384e-06,
+      "loss": 0.4854,
+      "step": 45460
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.459671166066869e-06,
+      "loss": 0.4906,
+      "step": 45480
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.45267099703849e-06,
+      "loss": 0.5068,
+      "step": 45500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.445679410780638e-06,
+      "loss": 0.5006,
+      "step": 45520
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4386964152441016e-06,
+      "loss": 0.4426,
+      "step": 45540
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.431722018369894e-06,
+      "loss": 0.5048,
+      "step": 45560
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.424756228089251e-06,
+      "loss": 0.4887,
+      "step": 45580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4177990523236216e-06,
+      "loss": 0.4839,
+      "step": 45600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.410850498984659e-06,
+      "loss": 0.4907,
+      "step": 45620
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4039105759742098e-06,
+      "loss": 0.5128,
+      "step": 45640
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.396979291184306e-06,
+      "loss": 0.5183,
+      "step": 45660
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3900566524971593e-06,
+      "loss": 0.4374,
+      "step": 45680
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3831426677851423e-06,
+      "loss": 0.5259,
+      "step": 45700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3762373449107935e-06,
+      "loss": 0.4733,
+      "step": 45720
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.369340691726798e-06,
+      "loss": 0.4627,
+      "step": 45740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3624527160759812e-06,
+      "loss": 0.4745,
+      "step": 45760
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3555734257913045e-06,
+      "loss": 0.4938,
+      "step": 45780
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3487028286958458e-06,
+      "loss": 0.4779,
+      "step": 45800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3418409326028037e-06,
+      "loss": 0.5329,
+      "step": 45820
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.334987745315478e-06,
+      "loss": 0.5157,
+      "step": 45840
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.328143274627266e-06,
+      "loss": 0.4395,
+      "step": 45860
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3213075283216547e-06,
+      "loss": 0.4928,
+      "step": 45880
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3144805141722038e-06,
+      "loss": 0.4894,
+      "step": 45900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3076622399425485e-06,
+      "loss": 0.486,
+      "step": 45920
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3008527133863836e-06,
+      "loss": 0.4598,
+      "step": 45940
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.294051942247457e-06,
+      "loss": 0.4787,
+      "step": 45960
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2872599342595557e-06,
+      "loss": 0.5415,
+      "step": 45980
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2804766971465054e-06,
+      "loss": 0.5304,
+      "step": 46000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2737022386221552e-06,
+      "loss": 0.5048,
+      "step": 46020
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2669365663903764e-06,
+      "loss": 0.5102,
+      "step": 46040
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2601796881450468e-06,
+      "loss": 0.5122,
+      "step": 46060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.253431611570035e-06,
+      "loss": 0.4871,
+      "step": 46080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.24669234433921e-06,
+      "loss": 0.465,
+      "step": 46100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2399618941164202e-06,
+      "loss": 0.4898,
+      "step": 46120
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2332402685554876e-06,
+      "loss": 0.4535,
+      "step": 46140
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2265274753002008e-06,
+      "loss": 0.4957,
+      "step": 46160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2198235219842967e-06,
+      "loss": 0.5043,
+      "step": 46180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.213128416231468e-06,
+      "loss": 0.4956,
+      "step": 46200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2064421656553415e-06,
+      "loss": 0.474,
+      "step": 46220
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.199764777859477e-06,
+      "loss": 0.5202,
+      "step": 46240
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1930962604373517e-06,
+      "loss": 0.4743,
+      "step": 46260
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1864366209723597e-06,
+      "loss": 0.5217,
+      "step": 46280
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.179785867037797e-06,
+      "loss": 0.544,
+      "step": 46300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1731440061968536e-06,
+      "loss": 0.5203,
+      "step": 46320
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.166511046002612e-06,
+      "loss": 0.4537,
+      "step": 46340
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1598869939980225e-06,
+      "loss": 0.5019,
+      "step": 46360
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.153271857715915e-06,
+      "loss": 0.4725,
+      "step": 46380
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1466656446789757e-06,
+      "loss": 0.5213,
+      "step": 46400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.140068362399745e-06,
+      "loss": 0.4668,
+      "step": 46420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.133480018380608e-06,
+      "loss": 0.4751,
+      "step": 46440
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1269006201137833e-06,
+      "loss": 0.4841,
+      "step": 46460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1203301750813166e-06,
+      "loss": 0.4737,
+      "step": 46480
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1137686907550737e-06,
+      "loss": 0.4682,
+      "step": 46500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1072161745967313e-06,
+      "loss": 0.516,
+      "step": 46520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1006726340577597e-06,
+      "loss": 0.4704,
+      "step": 46540
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.094138076579433e-06,
+      "loss": 0.4632,
+      "step": 46560
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.0876125095928033e-06,
+      "loss": 0.4756,
+      "step": 46580
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.0810959405187005e-06,
+      "loss": 0.4814,
+      "step": 46600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.0745883767677246e-06,
+      "loss": 0.5147,
+      "step": 46620
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.068089825740225e-06,
+      "loss": 0.489,
+      "step": 46640
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.0616002948263157e-06,
+      "loss": 0.4748,
+      "step": 46660
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.0551197914058464e-06,
+      "loss": 0.4802,
+      "step": 46680
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0486483228484012e-06,
+      "loss": 0.494,
+      "step": 46700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0421858965132856e-06,
+      "loss": 0.4977,
+      "step": 46720
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.035732519749528e-06,
+      "loss": 0.4961,
+      "step": 46740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0292881998958634e-06,
+      "loss": 0.4755,
+      "step": 46760
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.022852944280731e-06,
+      "loss": 0.4676,
+      "step": 46780
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.016426760222259e-06,
+      "loss": 0.4617,
+      "step": 46800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0100096550282543e-06,
+      "loss": 0.4792,
+      "step": 46820
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0036016359962084e-06,
+      "loss": 0.4592,
+      "step": 46840
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9972027104132763e-06,
+      "loss": 0.4831,
+      "step": 46860
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.990812885556269e-06,
+      "loss": 0.495,
+      "step": 46880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9844321686916602e-06,
+      "loss": 0.5107,
+      "step": 46900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9780605670755473e-06,
+      "loss": 0.4834,
+      "step": 46920
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.971698087953676e-06,
+      "loss": 0.5159,
+      "step": 46940
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.965344738561414e-06,
+      "loss": 0.4884,
+      "step": 46960
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9590005261237475e-06,
+      "loss": 0.4492,
+      "step": 46980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9526654578552673e-06,
+      "loss": 0.4833,
+      "step": 47000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9463395409601713e-06,
+      "loss": 0.4534,
+      "step": 47020
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.940022782632248e-06,
+      "loss": 0.4993,
+      "step": 47040
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9337151900548734e-06,
+      "loss": 0.4624,
+      "step": 47060
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9274167704009973e-06,
+      "loss": 0.481,
+      "step": 47080
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.921127530833138e-06,
+      "loss": 0.5272,
+      "step": 47100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9148474785033768e-06,
+      "loss": 0.4598,
+      "step": 47120
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.908576620553345e-06,
+      "loss": 0.4906,
+      "step": 47140
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.902314964114219e-06,
+      "loss": 0.4916,
+      "step": 47160
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8960625163067158e-06,
+      "loss": 0.5078,
+      "step": 47180
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.889819284241069e-06,
+      "loss": 0.4725,
+      "step": 47200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8835852750170414e-06,
+      "loss": 0.4765,
+      "step": 47220
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8773604957239067e-06,
+      "loss": 0.4832,
+      "step": 47240
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8711449534404424e-06,
+      "loss": 0.5203,
+      "step": 47260
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8649386552349136e-06,
+      "loss": 0.472,
+      "step": 47280
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8587416081650866e-06,
+      "loss": 0.4643,
+      "step": 47300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8525538192782e-06,
+      "loss": 0.4834,
+      "step": 47320
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8463752956109616e-06,
+      "loss": 0.5216,
+      "step": 47340
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8402060441895508e-06,
+      "loss": 0.5254,
+      "step": 47360
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8340460720295917e-06,
+      "loss": 0.4958,
+      "step": 47380
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.827895386136166e-06,
+      "loss": 0.4614,
+      "step": 47400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8217539935037897e-06,
+      "loss": 0.5094,
+      "step": 47420
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8156219011164121e-06,
+      "loss": 0.4292,
+      "step": 47440
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8094991159474107e-06,
+      "loss": 0.4841,
+      "step": 47460
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8033856449595689e-06,
+      "loss": 0.4885,
+      "step": 47480
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7972814951050855e-06,
+      "loss": 0.4551,
+      "step": 47500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7911866733255556e-06,
+      "loss": 0.4634,
+      "step": 47520
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.785101186551975e-06,
+      "loss": 0.4911,
+      "step": 47540
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.7790250417047116e-06,
+      "loss": 0.498,
+      "step": 47560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.772958245693518e-06,
+      "loss": 0.468,
+      "step": 47580
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7669008054175108e-06,
+      "loss": 0.4805,
+      "step": 47600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7608527277651711e-06,
+      "loss": 0.5318,
+      "step": 47620
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7548140196143338e-06,
+      "loss": 0.4472,
+      "step": 47640
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7487846878321691e-06,
+      "loss": 0.4444,
+      "step": 47660
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7427647392751967e-06,
+      "loss": 0.4735,
+      "step": 47680
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7367541807892586e-06,
+      "loss": 0.4689,
+      "step": 47700
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7307530192095212e-06,
+      "loss": 0.4587,
+      "step": 47720
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7247612613604637e-06,
+      "loss": 0.5297,
+      "step": 47740
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.718778914055873e-06,
+      "loss": 0.5013,
+      "step": 47760
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.712805984098832e-06,
+      "loss": 0.5003,
+      "step": 47780
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7068424782817162e-06,
+      "loss": 0.5118,
+      "step": 47800
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.700888403386186e-06,
+      "loss": 0.4921,
+      "step": 47820
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.694943766183168e-06,
+      "loss": 0.4948,
+      "step": 47840
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6890085734328666e-06,
+      "loss": 0.4913,
+      "step": 47860
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6830828318847415e-06,
+      "loss": 0.4905,
+      "step": 47880
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6771665482775058e-06,
+      "loss": 0.456,
+      "step": 47900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6712597293391153e-06,
+      "loss": 0.4673,
+      "step": 47920
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6653623817867638e-06,
+      "loss": 0.4698,
+      "step": 47940
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6594745123268751e-06,
+      "loss": 0.4524,
+      "step": 47960
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6535961276550948e-06,
+      "loss": 0.4678,
+      "step": 47980
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6477272344562789e-06,
+      "loss": 0.4924,
+      "step": 48000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6418678394044962e-06,
+      "loss": 0.4252,
+      "step": 48020
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6360179491630056e-06,
+      "loss": 0.4911,
+      "step": 48040
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6301775703842637e-06,
+      "loss": 0.4731,
+      "step": 48060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6243467097099097e-06,
+      "loss": 0.4915,
+      "step": 48080
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6185253737707607e-06,
+      "loss": 0.4807,
+      "step": 48100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6127135691867945e-06,
+      "loss": 0.4835,
+      "step": 48120
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6069113025671546e-06,
+      "loss": 0.5011,
+      "step": 48140
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6011185805101447e-06,
+      "loss": 0.5102,
+      "step": 48160
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5953354096032048e-06,
+      "loss": 0.4949,
+      "step": 48180
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5895617964229215e-06,
+      "loss": 0.4906,
+      "step": 48200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5837977475350008e-06,
+      "loss": 0.5041,
+      "step": 48220
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5780432694942815e-06,
+      "loss": 0.4643,
+      "step": 48240
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.572298368844719e-06,
+      "loss": 0.4583,
+      "step": 48260
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5665630521193753e-06,
+      "loss": 0.4727,
+      "step": 48280
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5608373258404087e-06,
+      "loss": 0.4544,
+      "step": 48300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5551211965190783e-06,
+      "loss": 0.5058,
+      "step": 48320
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5494146706557257e-06,
+      "loss": 0.4836,
+      "step": 48340
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5437177547397742e-06,
+      "loss": 0.4704,
+      "step": 48360
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.538030455249716e-06,
+      "loss": 0.4754,
+      "step": 48380
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5323527786531113e-06,
+      "loss": 0.5275,
+      "step": 48400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5266847314065726e-06,
+      "loss": 0.4518,
+      "step": 48420
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.521026319955764e-06,
+      "loss": 0.4905,
+      "step": 48440
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.515377550735394e-06,
+      "loss": 0.4558,
+      "step": 48460
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5097384301692041e-06,
+      "loss": 0.4613,
+      "step": 48480
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5041089646699603e-06,
+      "loss": 0.4816,
+      "step": 48500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4984891606394535e-06,
+      "loss": 0.4806,
+      "step": 48520
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4928790244684866e-06,
+      "loss": 0.4652,
+      "step": 48540
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.487278562536868e-06,
+      "loss": 0.492,
+      "step": 48560
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4816877812134034e-06,
+      "loss": 0.4711,
+      "step": 48580
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4761066868558915e-06,
+      "loss": 0.5062,
+      "step": 48600
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4705352858111144e-06,
+      "loss": 0.4772,
+      "step": 48620
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4649735844148304e-06,
+      "loss": 0.4634,
+      "step": 48640
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4594215889917707e-06,
+      "loss": 0.4814,
+      "step": 48660
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4538793058556222e-06,
+      "loss": 0.4659,
+      "step": 48680
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4483467413090324e-06,
+      "loss": 0.5063,
+      "step": 48700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4428239016435953e-06,
+      "loss": 0.4821,
+      "step": 48720
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.4373107931398477e-06,
+      "loss": 0.5402,
+      "step": 48740
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4318074220672595e-06,
+      "loss": 0.4746,
+      "step": 48760
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.426313794684221e-06,
+      "loss": 0.4944,
+      "step": 48780
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4208299172380512e-06,
+      "loss": 0.4909,
+      "step": 48800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4153557959649788e-06,
+      "loss": 0.4646,
+      "step": 48820
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4098914370901385e-06,
+      "loss": 0.4884,
+      "step": 48840
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4044368468275549e-06,
+      "loss": 0.4764,
+      "step": 48860
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3989920313801552e-06,
+      "loss": 0.5018,
+      "step": 48880
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3935569969397434e-06,
+      "loss": 0.4802,
+      "step": 48900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3881317496870062e-06,
+      "loss": 0.4864,
+      "step": 48920
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3827162957914975e-06,
+      "loss": 0.4689,
+      "step": 48940
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.37731064141163e-06,
+      "loss": 0.4675,
+      "step": 48960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3719147926946797e-06,
+      "loss": 0.4988,
+      "step": 48980
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3665287557767682e-06,
+      "loss": 0.4974,
+      "step": 49000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.361152536782856e-06,
+      "loss": 0.4712,
+      "step": 49020
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3557861418267514e-06,
+      "loss": 0.4812,
+      "step": 49040
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3504295770110731e-06,
+      "loss": 0.5029,
+      "step": 49060
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3450828484272727e-06,
+      "loss": 0.4581,
+      "step": 49080
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.4911,
+      "step": 49100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3344189242651672e-06,
+      "loss": 0.4557,
+      "step": 49120
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3291017408138007e-06,
+      "loss": 0.5026,
+      "step": 49140
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.32379441784818e-06,
+      "loss": 0.4814,
+      "step": 49160
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3184969614037567e-06,
+      "loss": 0.5167,
+      "step": 49180
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3132093775047616e-06,
+      "loss": 0.4855,
+      "step": 49200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3079316721641987e-06,
+      "loss": 0.4367,
+      "step": 49220
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3026638513838386e-06,
+      "loss": 0.4777,
+      "step": 49240
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.29740592115421e-06,
+      "loss": 0.4559,
+      "step": 49260
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.292157887454596e-06,
+      "loss": 0.4557,
+      "step": 49280
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2869197562530223e-06,
+      "loss": 0.5089,
+      "step": 49300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2816915335062596e-06,
+      "loss": 0.4675,
+      "step": 49320
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.2764732251598022e-06,
+      "loss": 0.4365,
+      "step": 49340
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2712648371478763e-06,
+      "loss": 0.4803,
+      "step": 49360
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2660663753934256e-06,
+      "loss": 0.507,
+      "step": 49380
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.260877845808105e-06,
+      "loss": 0.516,
+      "step": 49400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2556992542922708e-06,
+      "loss": 0.4704,
+      "step": 49420
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2505306067349853e-06,
+      "loss": 0.4862,
+      "step": 49440
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2453719090139994e-06,
+      "loss": 0.4913,
+      "step": 49460
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2402231669957465e-06,
+      "loss": 0.4875,
+      "step": 49480
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2350843865353447e-06,
+      "loss": 0.4714,
+      "step": 49500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2299555734765733e-06,
+      "loss": 0.4736,
+      "step": 49520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2248367336518874e-06,
+      "loss": 0.5141,
+      "step": 49540
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2197278728823947e-06,
+      "loss": 0.4822,
+      "step": 49560
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.214628996977857e-06,
+      "loss": 0.494,
+      "step": 49580
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2095401117366823e-06,
+      "loss": 0.5034,
+      "step": 49600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.204461222945913e-06,
+      "loss": 0.4749,
+      "step": 49620
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.1993923363812288e-06,
+      "loss": 0.4841,
+      "step": 49640
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1943334578069277e-06,
+      "loss": 0.4757,
+      "step": 49660
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1892845929759412e-06,
+      "loss": 0.4889,
+      "step": 49680
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1842457476297953e-06,
+      "loss": 0.4914,
+      "step": 49700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.179216927498632e-06,
+      "loss": 0.4723,
+      "step": 49720
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1741981383011924e-06,
+      "loss": 0.4435,
+      "step": 49740
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1691893857448089e-06,
+      "loss": 0.5092,
+      "step": 49760
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1641906755254007e-06,
+      "loss": 0.4701,
+      "step": 49780
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.159202013327464e-06,
+      "loss": 0.4724,
+      "step": 49800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.154223404824073e-06,
+      "loss": 0.4701,
+      "step": 49820
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1492548556768668e-06,
+      "loss": 0.4989,
+      "step": 49840
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1442963715360444e-06,
+      "loss": 0.4859,
+      "step": 49860
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.139347958040361e-06,
+      "loss": 0.4877,
+      "step": 49880
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1344096208171186e-06,
+      "loss": 0.5052,
+      "step": 49900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1294813654821591e-06,
+      "loss": 0.507,
+      "step": 49920
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1245631976398608e-06,
+      "loss": 0.5182,
+      "step": 49940
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1196551228831321e-06,
+      "loss": 0.4267,
+      "step": 49960
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1147571467933983e-06,
+      "loss": 0.4479,
+      "step": 49980
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1098692749406058e-06,
+      "loss": 0.4837,
+      "step": 50000
+    }
+  ],
+  "max_steps": 58920,
+  "num_train_epochs": 2,
+  "total_flos": 136141580967936.0,
+  "trial_name": null,
+  "trial_params": null
+}