diff --git "a/trainer_log_history.jsonl" "b/trainer_log_history.jsonl"
new file mode 100644--- /dev/null
+++ "b/trainer_log_history.jsonl"
@@ -0,0 +1,5397 @@
+{"loss": 2.1657, "learning_rate": 2.6999999999999996e-05, "epoch": 0.0, "step": 10}
+{"loss": 2.0891, "learning_rate": 5.6999999999999996e-05, "epoch": 0.0, "step": 20}
+{"loss": 1.8828, "learning_rate": 8.699999999999999e-05, "epoch": 0.0, "step": 30}
+{"loss": 1.5145, "learning_rate": 0.000117, "epoch": 0.0, "step": 40}
+{"loss": 1.1219, "learning_rate": 0.000147, "epoch": 0.0, "step": 50}
+{"loss": 1.05, "learning_rate": 0.00017699999999999997, "epoch": 0.0, "step": 60}
+{"loss": 0.8439, "learning_rate": 0.00020699999999999996, "epoch": 0.0, "step": 70}
+{"loss": 0.7854, "learning_rate": 0.000237, "epoch": 0.0, "step": 80}
+{"loss": 0.8191, "learning_rate": 0.000267, "epoch": 0.0, "step": 90}
+{"loss": 0.8437, "learning_rate": 0.00029699999999999996, "epoch": 0.0, "step": 100}
+{"loss": 0.8548, "learning_rate": 0.0002999498718948424, "epoch": 0.0, "step": 110}
+{"loss": 0.7763, "learning_rate": 0.0002998941740002228, "epoch": 0.0, "step": 120}
+{"loss": 0.751, "learning_rate": 0.0002998384761056032, "epoch": 0.0, "step": 130}
+{"loss": 0.8973, "learning_rate": 0.0002997827782109836, "epoch": 0.01, "step": 140}
+{"loss": 0.7949, "learning_rate": 0.00029972708031636403, "epoch": 0.01, "step": 150}
+{"loss": 0.7542, "learning_rate": 0.00029967138242174444, "epoch": 0.01, "step": 160}
+{"loss": 0.7928, "learning_rate": 0.00029961568452712485, "epoch": 0.01, "step": 170}
+{"loss": 1.0169, "learning_rate": 0.00029955998663250527, "epoch": 0.01, "step": 180}
+{"loss": 0.8765, "learning_rate": 0.0002995042887378857, "epoch": 0.01, "step": 190}
+{"loss": 0.8155, "learning_rate": 0.0002994485908432661, "epoch": 0.01, "step": 200}
+{"loss": 0.7217, "learning_rate": 0.0002993928929486465, "epoch": 0.01, "step": 210}
+{"loss": 0.8845, "learning_rate": 0.0002993371950540269, "epoch": 0.01, "step": 220}
+{"loss": 0.8273, "learning_rate": 0.0002992814971594073, "epoch": 0.01, "step": 230}
+{"loss": 0.7669, "learning_rate": 0.0002992257992647878, "epoch": 0.01, "step": 240}
+{"loss": 0.8067, "learning_rate": 0.0002991701013701682, "epoch": 0.01, "step": 250}
+{"loss": 0.7247, "learning_rate": 0.0002991144034755486, "epoch": 0.01, "step": 260}
+{"loss": 0.7256, "learning_rate": 0.000299058705580929, "epoch": 0.01, "step": 270}
+{"loss": 0.8089, "learning_rate": 0.00029900300768630943, "epoch": 0.01, "step": 280}
+{"loss": 0.714, "learning_rate": 0.00029894730979168984, "epoch": 0.01, "step": 290}
+{"loss": 0.852, "learning_rate": 0.00029889161189707025, "epoch": 0.01, "step": 300}
+{"loss": 0.8713, "learning_rate": 0.0002988359140024507, "epoch": 0.01, "step": 310}
+{"loss": 0.9165, "learning_rate": 0.00029878021610783113, "epoch": 0.01, "step": 320}
+{"loss": 0.8829, "learning_rate": 0.00029872451821321154, "epoch": 0.01, "step": 330}
+{"loss": 0.8993, "learning_rate": 0.00029866882031859195, "epoch": 0.01, "step": 340}
+{"loss": 0.7899, "learning_rate": 0.00029861312242397236, "epoch": 0.01, "step": 350}
+{"loss": 0.8127, "learning_rate": 0.0002985574245293528, "epoch": 0.01, "step": 360}
+{"loss": 0.7791, "learning_rate": 0.0002985017266347332, "epoch": 0.01, "step": 370}
+{"loss": 0.801, "learning_rate": 0.0002984460287401136, "epoch": 0.01, "step": 380}
+{"loss": 0.9644, "learning_rate": 0.000298390330845494, "epoch": 0.01, "step": 390}
+{"loss": 0.8106, "learning_rate": 0.0002983346329508744, "epoch": 0.01, "step": 400}
+{"loss": 0.7146, "learning_rate": 0.00029827893505625483, "epoch": 0.02, "step": 410}
+{"loss": 0.809, "learning_rate": 0.00029822323716163524, "epoch": 0.02, "step": 420}
+{"loss": 0.7411, "learning_rate": 0.00029816753926701565, "epoch": 0.02, "step": 430}
+{"loss": 0.8484, "learning_rate": 0.0002981118413723961, "epoch": 0.02, "step": 440}
+{"loss": 0.8111, "learning_rate": 0.00029805614347777653, "epoch": 0.02, "step": 450}
+{"loss": 0.8426, "learning_rate": 0.00029800044558315694, "epoch": 0.02, "step": 460}
+{"loss": 0.8323, "learning_rate": 0.00029794474768853735, "epoch": 0.02, "step": 470}
+{"loss": 0.8651, "learning_rate": 0.00029788904979391776, "epoch": 0.02, "step": 480}
+{"loss": 0.7317, "learning_rate": 0.0002978333518992982, "epoch": 0.02, "step": 490}
+{"loss": 0.8968, "learning_rate": 0.00029777765400467864, "epoch": 0.02, "step": 500}
+{"loss": 0.8402, "learning_rate": 0.00029772195611005905, "epoch": 0.02, "step": 510}
+{"loss": 0.8242, "learning_rate": 0.00029766625821543946, "epoch": 0.02, "step": 520}
+{"loss": 0.8364, "learning_rate": 0.00029761056032081987, "epoch": 0.02, "step": 530}
+{"loss": 0.6825, "learning_rate": 0.0002975548624262003, "epoch": 0.02, "step": 540}
+{"loss": 0.7819, "learning_rate": 0.0002974991645315807, "epoch": 0.02, "step": 550}
+{"loss": 0.7209, "learning_rate": 0.0002974434666369611, "epoch": 0.02, "step": 560}
+{"loss": 0.6985, "learning_rate": 0.0002973877687423415, "epoch": 0.02, "step": 570}
+{"loss": 0.8092, "learning_rate": 0.00029733207084772193, "epoch": 0.02, "step": 580}
+{"loss": 0.7812, "learning_rate": 0.00029727637295310234, "epoch": 0.02, "step": 590}
+{"loss": 0.8383, "learning_rate": 0.00029722067505848275, "epoch": 0.02, "step": 600}
+{"loss": 0.676, "learning_rate": 0.00029716497716386316, "epoch": 0.02, "step": 610}
+{"loss": 0.8265, "learning_rate": 0.00029710927926924357, "epoch": 0.02, "step": 620}
+{"loss": 0.8141, "learning_rate": 0.00029705358137462404, "epoch": 0.02, "step": 630}
+{"loss": 0.9589, "learning_rate": 0.00029699788348000445, "epoch": 0.02, "step": 640}
+{"loss": 0.7259, "learning_rate": 0.00029694218558538486, "epoch": 0.02, "step": 650}
+{"loss": 0.9564, "learning_rate": 0.00029688648769076527, "epoch": 0.02, "step": 660}
+{"loss": 0.7736, "learning_rate": 0.0002968307897961457, "epoch": 0.02, "step": 670}
+{"loss": 0.8092, "learning_rate": 0.0002967750919015261, "epoch": 0.03, "step": 680}
+{"loss": 0.6447, "learning_rate": 0.0002967193940069065, "epoch": 0.03, "step": 690}
+{"loss": 0.797, "learning_rate": 0.0002966636961122869, "epoch": 0.03, "step": 700}
+{"loss": 0.8556, "learning_rate": 0.0002966079982176673, "epoch": 0.03, "step": 710}
+{"loss": 0.7671, "learning_rate": 0.00029655230032304774, "epoch": 0.03, "step": 720}
+{"loss": 0.7346, "learning_rate": 0.0002964966024284282, "epoch": 0.03, "step": 730}
+{"loss": 0.8341, "learning_rate": 0.0002964409045338086, "epoch": 0.03, "step": 740}
+{"loss": 0.7651, "learning_rate": 0.000296385206639189, "epoch": 0.03, "step": 750}
+{"loss": 0.6583, "learning_rate": 0.00029632950874456944, "epoch": 0.03, "step": 760}
+{"loss": 0.9385, "learning_rate": 0.00029627381084994985, "epoch": 0.03, "step": 770}
+{"loss": 0.7709, "learning_rate": 0.00029621811295533026, "epoch": 0.03, "step": 780}
+{"loss": 0.7106, "learning_rate": 0.00029616241506071067, "epoch": 0.03, "step": 790}
+{"loss": 0.8967, "learning_rate": 0.0002961067171660911, "epoch": 0.03, "step": 800}
+{"loss": 0.7392, "learning_rate": 0.0002960510192714715, "epoch": 0.03, "step": 810}
+{"loss": 0.7959, "learning_rate": 0.0002959953213768519, "epoch": 0.03, "step": 820}
+{"loss": 0.8163, "learning_rate": 0.00029593962348223237, "epoch": 0.03, "step": 830}
+{"loss": 0.886, "learning_rate": 0.0002958839255876128, "epoch": 0.03, "step": 840}
+{"loss": 0.7094, "learning_rate": 0.0002958282276929932, "epoch": 0.03, "step": 850}
+{"loss": 0.7543, "learning_rate": 0.0002957725297983736, "epoch": 0.03, "step": 860}
+{"loss": 0.7067, "learning_rate": 0.000295716831903754, "epoch": 0.03, "step": 870}
+{"loss": 0.7805, "learning_rate": 0.0002956611340091344, "epoch": 0.03, "step": 880}
+{"loss": 0.7247, "learning_rate": 0.00029560543611451484, "epoch": 0.03, "step": 890}
+{"loss": 0.748, "learning_rate": 0.00029554973821989525, "epoch": 0.03, "step": 900}
+{"loss": 0.8429, "learning_rate": 0.00029549404032527566, "epoch": 0.03, "step": 910}
+{"loss": 0.7323, "learning_rate": 0.00029543834243065607, "epoch": 0.03, "step": 920}
+{"loss": 0.6964, "learning_rate": 0.0002953826445360365, "epoch": 0.03, "step": 930}
+{"loss": 0.7016, "learning_rate": 0.0002953269466414169, "epoch": 0.03, "step": 940}
+{"loss": 0.9425, "learning_rate": 0.0002952712487467973, "epoch": 0.04, "step": 950}
+{"loss": 0.7805, "learning_rate": 0.00029521555085217777, "epoch": 0.04, "step": 960}
+{"loss": 0.7907, "learning_rate": 0.0002951598529575582, "epoch": 0.04, "step": 970}
+{"loss": 0.6496, "learning_rate": 0.0002951041550629386, "epoch": 0.04, "step": 980}
+{"loss": 0.8093, "learning_rate": 0.000295048457168319, "epoch": 0.04, "step": 990}
+{"loss": 0.7786, "learning_rate": 0.0002949927592736994, "epoch": 0.04, "step": 1000}
+{"loss": 0.8045, "learning_rate": 0.0002949370613790798, "epoch": 0.04, "step": 1010}
+{"loss": 0.7455, "learning_rate": 0.0002948813634844603, "epoch": 0.04, "step": 1020}
+{"loss": 0.8247, "learning_rate": 0.0002948256655898407, "epoch": 0.04, "step": 1030}
+{"loss": 0.732, "learning_rate": 0.0002947699676952211, "epoch": 0.04, "step": 1040}
+{"loss": 0.7978, "learning_rate": 0.0002947142698006015, "epoch": 0.04, "step": 1050}
+{"loss": 0.8016, "learning_rate": 0.00029465857190598193, "epoch": 0.04, "step": 1060}
+{"loss": 0.7567, "learning_rate": 0.00029460287401136234, "epoch": 0.04, "step": 1070}
+{"loss": 0.5406, "learning_rate": 0.00029454717611674276, "epoch": 0.04, "step": 1080}
+{"loss": 0.9177, "learning_rate": 0.00029449147822212317, "epoch": 0.04, "step": 1090}
+{"loss": 0.7351, "learning_rate": 0.0002944357803275036, "epoch": 0.04, "step": 1100}
+{"loss": 0.7508, "learning_rate": 0.000294380082432884, "epoch": 0.04, "step": 1110}
+{"loss": 0.7642, "learning_rate": 0.0002943243845382644, "epoch": 0.04, "step": 1120}
+{"loss": 0.7415, "learning_rate": 0.0002942686866436448, "epoch": 0.04, "step": 1130}
+{"loss": 0.8158, "learning_rate": 0.0002942129887490252, "epoch": 0.04, "step": 1140}
+{"loss": 0.8524, "learning_rate": 0.0002941572908544057, "epoch": 0.04, "step": 1150}
+{"loss": 0.741, "learning_rate": 0.0002941015929597861, "epoch": 0.04, "step": 1160}
+{"loss": 0.6836, "learning_rate": 0.0002940458950651665, "epoch": 0.04, "step": 1170}
+{"loss": 0.6417, "learning_rate": 0.0002939901971705469, "epoch": 0.04, "step": 1180}
+{"loss": 0.8078, "learning_rate": 0.00029393449927592733, "epoch": 0.04, "step": 1190}
+{"loss": 0.6682, "learning_rate": 0.00029387880138130774, "epoch": 0.04, "step": 1200}
+{"loss": 0.8078, "learning_rate": 0.0002938231034866882, "epoch": 0.04, "step": 1210}
+{"loss": 0.7706, "learning_rate": 0.0002937674055920686, "epoch": 0.05, "step": 1220}
+{"loss": 0.7169, "learning_rate": 0.00029371170769744903, "epoch": 0.05, "step": 1230}
+{"loss": 0.8704, "learning_rate": 0.00029365600980282944, "epoch": 0.05, "step": 1240}
+{"loss": 0.8781, "learning_rate": 0.00029360031190820985, "epoch": 0.05, "step": 1250}
+{"loss": 0.7668, "learning_rate": 0.00029354461401359026, "epoch": 0.05, "step": 1260}
+{"loss": 0.9439, "learning_rate": 0.0002934889161189707, "epoch": 0.05, "step": 1270}
+{"loss": 0.7886, "learning_rate": 0.0002934332182243511, "epoch": 0.05, "step": 1280}
+{"loss": 0.6752, "learning_rate": 0.0002933775203297315, "epoch": 0.05, "step": 1290}
+{"loss": 0.9126, "learning_rate": 0.0002933218224351119, "epoch": 0.05, "step": 1300}
+{"loss": 0.7678, "learning_rate": 0.0002932661245404923, "epoch": 0.05, "step": 1310}
+{"loss": 0.7099, "learning_rate": 0.00029321042664587273, "epoch": 0.05, "step": 1320}
+{"loss": 0.7767, "learning_rate": 0.00029315472875125314, "epoch": 0.05, "step": 1330}
+{"loss": 0.8892, "learning_rate": 0.0002930990308566336, "epoch": 0.05, "step": 1340}
+{"loss": 0.7937, "learning_rate": 0.000293043332962014, "epoch": 0.05, "step": 1350}
+{"loss": 0.6858, "learning_rate": 0.00029298763506739443, "epoch": 0.05, "step": 1360}
+{"loss": 0.7289, "learning_rate": 0.00029293193717277484, "epoch": 0.05, "step": 1370}
+{"loss": 0.747, "learning_rate": 0.00029287623927815525, "epoch": 0.05, "step": 1380}
+{"loss": 0.7755, "learning_rate": 0.00029282054138353566, "epoch": 0.05, "step": 1390}
+{"loss": 0.8012, "learning_rate": 0.0002927648434889161, "epoch": 0.05, "step": 1400}
+{"loss": 0.7436, "learning_rate": 0.00029270914559429654, "epoch": 0.05, "step": 1410}
+{"loss": 0.7274, "learning_rate": 0.00029265344769967695, "epoch": 0.05, "step": 1420}
+{"loss": 0.7739, "learning_rate": 0.00029259774980505736, "epoch": 0.05, "step": 1430}
+{"loss": 0.6398, "learning_rate": 0.0002925420519104378, "epoch": 0.05, "step": 1440}
+{"loss": 0.8671, "learning_rate": 0.0002924863540158182, "epoch": 0.05, "step": 1450}
+{"loss": 0.7438, "learning_rate": 0.0002924306561211986, "epoch": 0.05, "step": 1460}
+{"loss": 0.7123, "learning_rate": 0.000292374958226579, "epoch": 0.05, "step": 1470}
+{"loss": 0.7453, "learning_rate": 0.0002923192603319594, "epoch": 0.05, "step": 1480}
+{"loss": 0.7307, "learning_rate": 0.00029226356243733983, "epoch": 0.06, "step": 1490}
+{"loss": 0.8455, "learning_rate": 0.00029220786454272024, "epoch": 0.06, "step": 1500}
+{"loss": 0.7856, "learning_rate": 0.00029215216664810065, "epoch": 0.06, "step": 1510}
+{"loss": 0.8438, "learning_rate": 0.00029209646875348106, "epoch": 0.06, "step": 1520}
+{"loss": 0.7787, "learning_rate": 0.0002920407708588615, "epoch": 0.06, "step": 1530}
+{"loss": 0.7922, "learning_rate": 0.00029198507296424194, "epoch": 0.06, "step": 1540}
+{"loss": 0.7894, "learning_rate": 0.00029192937506962235, "epoch": 0.06, "step": 1550}
+{"loss": 0.6718, "learning_rate": 0.00029187367717500276, "epoch": 0.06, "step": 1560}
+{"loss": 0.7892, "learning_rate": 0.00029181797928038317, "epoch": 0.06, "step": 1570}
+{"loss": 1.0537, "learning_rate": 0.0002917622813857636, "epoch": 0.06, "step": 1580}
+{"loss": 0.823, "learning_rate": 0.000291706583491144, "epoch": 0.06, "step": 1590}
+{"loss": 0.6297, "learning_rate": 0.00029165088559652446, "epoch": 0.06, "step": 1600}
+{"loss": 0.6563, "learning_rate": 0.00029159518770190487, "epoch": 0.06, "step": 1610}
+{"loss": 0.6895, "learning_rate": 0.0002915394898072853, "epoch": 0.06, "step": 1620}
+{"loss": 0.7973, "learning_rate": 0.0002914837919126657, "epoch": 0.06, "step": 1630}
+{"loss": 0.7643, "learning_rate": 0.0002914280940180461, "epoch": 0.06, "step": 1640}
+{"loss": 0.8495, "learning_rate": 0.0002913723961234265, "epoch": 0.06, "step": 1650}
+{"loss": 0.9305, "learning_rate": 0.0002913166982288069, "epoch": 0.06, "step": 1660}
+{"loss": 0.768, "learning_rate": 0.00029126100033418734, "epoch": 0.06, "step": 1670}
+{"loss": 0.7529, "learning_rate": 0.00029120530243956775, "epoch": 0.06, "step": 1680}
+{"loss": 0.8489, "learning_rate": 0.00029114960454494816, "epoch": 0.06, "step": 1690}
+{"loss": 0.6501, "learning_rate": 0.00029109390665032857, "epoch": 0.06, "step": 1700}
+{"loss": 0.7523, "learning_rate": 0.000291038208755709, "epoch": 0.06, "step": 1710}
+{"loss": 0.7792, "learning_rate": 0.0002909825108610894, "epoch": 0.06, "step": 1720}
+{"loss": 0.7997, "learning_rate": 0.00029092681296646986, "epoch": 0.06, "step": 1730}
+{"loss": 0.8561, "learning_rate": 0.00029087111507185027, "epoch": 0.06, "step": 1740}
+{"loss": 0.8504, "learning_rate": 0.0002908154171772307, "epoch": 0.06, "step": 1750}
+{"loss": 0.7216, "learning_rate": 0.0002907597192826111, "epoch": 0.07, "step": 1760}
+{"loss": 0.7572, "learning_rate": 0.0002907040213879915, "epoch": 0.07, "step": 1770}
+{"loss": 0.6785, "learning_rate": 0.0002906483234933719, "epoch": 0.07, "step": 1780}
+{"loss": 0.8522, "learning_rate": 0.0002905926255987523, "epoch": 0.07, "step": 1790}
+{"loss": 0.622, "learning_rate": 0.0002905369277041328, "epoch": 0.07, "step": 1800}
+{"loss": 0.7532, "learning_rate": 0.0002904812298095132, "epoch": 0.07, "step": 1810}
+{"loss": 0.7477, "learning_rate": 0.0002904255319148936, "epoch": 0.07, "step": 1820}
+{"loss": 0.737, "learning_rate": 0.000290369834020274, "epoch": 0.07, "step": 1830}
+{"loss": 0.8347, "learning_rate": 0.00029031413612565444, "epoch": 0.07, "step": 1840}
+{"loss": 0.6236, "learning_rate": 0.00029025843823103485, "epoch": 0.07, "step": 1850}
+{"loss": 0.894, "learning_rate": 0.00029020274033641526, "epoch": 0.07, "step": 1860}
+{"loss": 0.829, "learning_rate": 0.00029014704244179567, "epoch": 0.07, "step": 1870}
+{"loss": 0.655, "learning_rate": 0.0002900913445471761, "epoch": 0.07, "step": 1880}
+{"loss": 0.8142, "learning_rate": 0.0002900356466525565, "epoch": 0.07, "step": 1890}
+{"loss": 0.8483, "learning_rate": 0.0002899799487579369, "epoch": 0.07, "step": 1900}
+{"loss": 0.8201, "learning_rate": 0.0002899242508633173, "epoch": 0.07, "step": 1910}
+{"loss": 0.6601, "learning_rate": 0.0002898685529686977, "epoch": 0.07, "step": 1920}
+{"loss": 0.8202, "learning_rate": 0.0002898128550740782, "epoch": 0.07, "step": 1930}
+{"loss": 0.6678, "learning_rate": 0.0002897571571794586, "epoch": 0.07, "step": 1940}
+{"loss": 0.7731, "learning_rate": 0.000289701459284839, "epoch": 0.07, "step": 1950}
+{"loss": 0.7277, "learning_rate": 0.0002896457613902194, "epoch": 0.07, "step": 1960}
+{"loss": 0.6507, "learning_rate": 0.00028959563328506184, "epoch": 0.07, "step": 1970}
+{"loss": 0.7655, "learning_rate": 0.00028953993539044225, "epoch": 0.07, "step": 1980}
+{"loss": 0.7873, "learning_rate": 0.00028948423749582266, "epoch": 0.07, "step": 1990}
+{"loss": 0.9066, "learning_rate": 0.00028942853960120307, "epoch": 0.07, "step": 2000}
+{"loss": 0.8656, "learning_rate": 0.0002893728417065835, "epoch": 0.07, "step": 2010}
+{"loss": 0.7119, "learning_rate": 0.0002893171438119639, "epoch": 0.07, "step": 2020}
+{"loss": 0.76, "learning_rate": 0.0002892614459173443, "epoch": 0.08, "step": 2030}
+{"loss": 0.659, "learning_rate": 0.0002892057480227247, "epoch": 0.08, "step": 2040}
+{"loss": 0.6439, "learning_rate": 0.0002891500501281051, "epoch": 0.08, "step": 2050}
+{"loss": 0.6981, "learning_rate": 0.00028909435223348554, "epoch": 0.08, "step": 2060}
+{"loss": 0.7457, "learning_rate": 0.00028903865433886595, "epoch": 0.08, "step": 2070}
+{"loss": 0.7324, "learning_rate": 0.00028898295644424636, "epoch": 0.08, "step": 2080}
+{"loss": 0.8091, "learning_rate": 0.00028892725854962677, "epoch": 0.08, "step": 2090}
+{"loss": 0.7322, "learning_rate": 0.00028887156065500724, "epoch": 0.08, "step": 2100}
+{"loss": 0.7112, "learning_rate": 0.00028881586276038765, "epoch": 0.08, "step": 2110}
+{"loss": 0.8487, "learning_rate": 0.00028876016486576806, "epoch": 0.08, "step": 2120}
+{"loss": 0.6818, "learning_rate": 0.00028870446697114847, "epoch": 0.08, "step": 2130}
+{"loss": 0.7988, "learning_rate": 0.0002886487690765289, "epoch": 0.08, "step": 2140}
+{"loss": 0.7921, "learning_rate": 0.0002885930711819093, "epoch": 0.08, "step": 2150}
+{"loss": 0.7544, "learning_rate": 0.00028853737328728976, "epoch": 0.08, "step": 2160}
+{"loss": 0.7076, "learning_rate": 0.00028848167539267017, "epoch": 0.08, "step": 2170}
+{"loss": 0.7264, "learning_rate": 0.0002884259774980506, "epoch": 0.08, "step": 2180}
+{"loss": 0.8993, "learning_rate": 0.000288370279603431, "epoch": 0.08, "step": 2190}
+{"loss": 0.7293, "learning_rate": 0.0002883145817088114, "epoch": 0.08, "step": 2200}
+{"loss": 0.7917, "learning_rate": 0.0002882588838141918, "epoch": 0.08, "step": 2210}
+{"loss": 0.6635, "learning_rate": 0.0002882031859195722, "epoch": 0.08, "step": 2220}
+{"loss": 0.7215, "learning_rate": 0.00028814748802495263, "epoch": 0.08, "step": 2230}
+{"loss": 0.7617, "learning_rate": 0.00028809179013033305, "epoch": 0.08, "step": 2240}
+{"loss": 0.8387, "learning_rate": 0.00028803609223571346, "epoch": 0.08, "step": 2250}
+{"loss": 0.665, "learning_rate": 0.00028798039434109387, "epoch": 0.08, "step": 2260}
+{"loss": 0.8129, "learning_rate": 0.0002879246964464743, "epoch": 0.08, "step": 2270}
+{"loss": 0.5688, "learning_rate": 0.0002878689985518547, "epoch": 0.08, "step": 2280}
+{"loss": 0.7847, "learning_rate": 0.00028781330065723516, "epoch": 0.08, "step": 2290}
+{"loss": 0.6735, "learning_rate": 0.00028775760276261557, "epoch": 0.09, "step": 2300}
+{"loss": 0.5909, "learning_rate": 0.000287701904867996, "epoch": 0.09, "step": 2310}
+{"loss": 0.788, "learning_rate": 0.0002876462069733764, "epoch": 0.09, "step": 2320}
+{"loss": 0.783, "learning_rate": 0.0002875905090787568, "epoch": 0.09, "step": 2330}
+{"loss": 0.7952, "learning_rate": 0.0002875348111841372, "epoch": 0.09, "step": 2340}
+{"loss": 0.6714, "learning_rate": 0.0002874791132895176, "epoch": 0.09, "step": 2350}
+{"loss": 0.7481, "learning_rate": 0.00028742341539489803, "epoch": 0.09, "step": 2360}
+{"loss": 0.6156, "learning_rate": 0.00028736771750027844, "epoch": 0.09, "step": 2370}
+{"loss": 0.7312, "learning_rate": 0.0002873120196056589, "epoch": 0.09, "step": 2380}
+{"loss": 0.7561, "learning_rate": 0.0002872563217110393, "epoch": 0.09, "step": 2390}
+{"loss": 0.7927, "learning_rate": 0.00028720062381641973, "epoch": 0.09, "step": 2400}
+{"loss": 0.7223, "learning_rate": 0.00028714492592180014, "epoch": 0.09, "step": 2410}
+{"loss": 0.7335, "learning_rate": 0.00028708922802718055, "epoch": 0.09, "step": 2420}
+{"loss": 0.6564, "learning_rate": 0.00028703353013256097, "epoch": 0.09, "step": 2430}
+{"loss": 0.6481, "learning_rate": 0.0002869778322379414, "epoch": 0.09, "step": 2440}
+{"loss": 0.9089, "learning_rate": 0.0002869221343433218, "epoch": 0.09, "step": 2450}
+{"loss": 0.7339, "learning_rate": 0.0002868664364487022, "epoch": 0.09, "step": 2460}
+{"loss": 0.6527, "learning_rate": 0.0002868107385540826, "epoch": 0.09, "step": 2470}
+{"loss": 0.6925, "learning_rate": 0.000286755040659463, "epoch": 0.09, "step": 2480}
+{"loss": 0.691, "learning_rate": 0.0002866993427648435, "epoch": 0.09, "step": 2490}
+{"loss": 0.6942, "learning_rate": 0.0002866436448702239, "epoch": 0.09, "step": 2500}
+{"loss": 0.8531, "learning_rate": 0.0002865879469756043, "epoch": 0.09, "step": 2510}
+{"loss": 0.9173, "learning_rate": 0.0002865322490809847, "epoch": 0.09, "step": 2520}
+{"loss": 0.7681, "learning_rate": 0.00028647655118636513, "epoch": 0.09, "step": 2530}
+{"loss": 0.7211, "learning_rate": 0.00028642085329174554, "epoch": 0.09, "step": 2540}
+{"loss": 0.6427, "learning_rate": 0.00028636515539712595, "epoch": 0.09, "step": 2550}
+{"loss": 0.7184, "learning_rate": 0.00028630945750250636, "epoch": 0.09, "step": 2560}
+{"loss": 0.6873, "learning_rate": 0.0002862537596078868, "epoch": 0.1, "step": 2570}
+{"loss": 0.6968, "learning_rate": 0.0002861980617132672, "epoch": 0.1, "step": 2580}
+{"loss": 0.788, "learning_rate": 0.0002861423638186476, "epoch": 0.1, "step": 2590}
+{"loss": 0.7222, "learning_rate": 0.000286086665924028, "epoch": 0.1, "step": 2600}
+{"loss": 0.8198, "learning_rate": 0.0002860309680294084, "epoch": 0.1, "step": 2610}
+{"loss": 0.5913, "learning_rate": 0.0002859752701347889, "epoch": 0.1, "step": 2620}
+{"loss": 0.686, "learning_rate": 0.0002859195722401693, "epoch": 0.1, "step": 2630}
+{"loss": 0.7338, "learning_rate": 0.0002858638743455497, "epoch": 0.1, "step": 2640}
+{"loss": 0.7901, "learning_rate": 0.0002858081764509301, "epoch": 0.1, "step": 2650}
+{"loss": 0.6926, "learning_rate": 0.00028575247855631053, "epoch": 0.1, "step": 2660}
+{"loss": 0.7756, "learning_rate": 0.00028569678066169094, "epoch": 0.1, "step": 2670}
+{"loss": 0.7851, "learning_rate": 0.0002856410827670714, "epoch": 0.1, "step": 2680}
+{"loss": 0.9493, "learning_rate": 0.0002855853848724518, "epoch": 0.1, "step": 2690}
+{"loss": 0.8183, "learning_rate": 0.00028552968697783223, "epoch": 0.1, "step": 2700}
+{"loss": 0.769, "learning_rate": 0.00028547398908321264, "epoch": 0.1, "step": 2710}
+{"loss": 0.7693, "learning_rate": 0.00028541829118859305, "epoch": 0.1, "step": 2720}
+{"loss": 0.649, "learning_rate": 0.00028536259329397346, "epoch": 0.1, "step": 2730}
+{"loss": 0.6816, "learning_rate": 0.0002853068953993539, "epoch": 0.1, "step": 2740}
+{"loss": 0.8367, "learning_rate": 0.0002852511975047343, "epoch": 0.1, "step": 2750}
+{"loss": 0.6146, "learning_rate": 0.0002851954996101147, "epoch": 0.1, "step": 2760}
+{"loss": 0.7475, "learning_rate": 0.0002851398017154951, "epoch": 0.1, "step": 2770}
+{"loss": 0.7175, "learning_rate": 0.0002850841038208755, "epoch": 0.1, "step": 2780}
+{"loss": 0.642, "learning_rate": 0.00028502840592625593, "epoch": 0.1, "step": 2790}
+{"loss": 0.7133, "learning_rate": 0.00028497270803163634, "epoch": 0.1, "step": 2800}
+{"loss": 0.5929, "learning_rate": 0.0002849170101370168, "epoch": 0.1, "step": 2810}
+{"loss": 0.7594, "learning_rate": 0.0002848613122423972, "epoch": 0.1, "step": 2820}
+{"loss": 0.7461, "learning_rate": 0.00028480561434777763, "epoch": 0.1, "step": 2830}
+{"loss": 0.7089, "learning_rate": 0.00028474991645315804, "epoch": 0.11, "step": 2840}
+{"loss": 0.805, "learning_rate": 0.00028469421855853845, "epoch": 0.11, "step": 2850}
+{"loss": 0.8134, "learning_rate": 0.00028463852066391886, "epoch": 0.11, "step": 2860}
+{"loss": 0.7319, "learning_rate": 0.00028458282276929927, "epoch": 0.11, "step": 2870}
+{"loss": 0.7903, "learning_rate": 0.00028452712487467974, "epoch": 0.11, "step": 2880}
+{"loss": 0.7304, "learning_rate": 0.00028447142698006015, "epoch": 0.11, "step": 2890}
+{"loss": 0.7112, "learning_rate": 0.00028441572908544056, "epoch": 0.11, "step": 2900}
+{"loss": 0.7919, "learning_rate": 0.00028436003119082097, "epoch": 0.11, "step": 2910}
+{"loss": 0.6882, "learning_rate": 0.0002843043332962014, "epoch": 0.11, "step": 2920}
+{"loss": 0.7139, "learning_rate": 0.0002842486354015818, "epoch": 0.11, "step": 2930}
+{"loss": 0.7818, "learning_rate": 0.0002841929375069622, "epoch": 0.11, "step": 2940}
+{"loss": 0.6862, "learning_rate": 0.0002841372396123426, "epoch": 0.11, "step": 2950}
+{"loss": 0.8077, "learning_rate": 0.000284081541717723, "epoch": 0.11, "step": 2960}
+{"loss": 0.8587, "learning_rate": 0.00028402584382310344, "epoch": 0.11, "step": 2970}
+{"loss": 0.6373, "learning_rate": 0.00028397014592848385, "epoch": 0.11, "step": 2980}
+{"loss": 0.7463, "learning_rate": 0.00028391444803386426, "epoch": 0.11, "step": 2990}
+{"loss": 0.7491, "learning_rate": 0.0002838587501392447, "epoch": 0.11, "step": 3000}
+{"loss": 0.8382, "learning_rate": 0.00028380305224462514, "epoch": 0.11, "step": 3010}
+{"loss": 0.7828, "learning_rate": 0.00028374735435000555, "epoch": 0.11, "step": 3020}
+{"loss": 0.8557, "learning_rate": 0.00028369165645538596, "epoch": 0.11, "step": 3030}
+{"loss": 0.723, "learning_rate": 0.00028363595856076637, "epoch": 0.11, "step": 3040}
+{"loss": 0.8736, "learning_rate": 0.0002835802606661468, "epoch": 0.11, "step": 3050}
+{"loss": 0.6935, "learning_rate": 0.0002835245627715272, "epoch": 0.11, "step": 3060}
+{"loss": 0.67, "learning_rate": 0.00028346886487690766, "epoch": 0.11, "step": 3070}
+{"loss": 0.8309, "learning_rate": 0.00028341316698228807, "epoch": 0.11, "step": 3080}
+{"loss": 0.7515, "learning_rate": 0.0002833574690876685, "epoch": 0.11, "step": 3090}
+{"loss": 0.7052, "learning_rate": 0.0002833017711930489, "epoch": 0.11, "step": 3100}
+{"loss": 0.8169, "learning_rate": 0.0002832460732984293, "epoch": 0.12, "step": 3110}
+{"loss": 0.6467, "learning_rate": 0.0002831903754038097, "epoch": 0.12, "step": 3120}
+{"loss": 0.6708, "learning_rate": 0.0002831346775091901, "epoch": 0.12, "step": 3130}
+{"loss": 0.7488, "learning_rate": 0.00028307897961457054, "epoch": 0.12, "step": 3140}
+{"loss": 0.6892, "learning_rate": 0.00028302328171995095, "epoch": 0.12, "step": 3150}
+{"loss": 0.7342, "learning_rate": 0.00028296758382533136, "epoch": 0.12, "step": 3160}
+{"loss": 0.6201, "learning_rate": 0.00028291188593071177, "epoch": 0.12, "step": 3170}
+{"loss": 0.881, "learning_rate": 0.0002828561880360922, "epoch": 0.12, "step": 3180}
+{"loss": 0.7862, "learning_rate": 0.0002828004901414726, "epoch": 0.12, "step": 3190}
+{"loss": 0.7323, "learning_rate": 0.00028274479224685306, "epoch": 0.12, "step": 3200}
+{"loss": 0.708, "learning_rate": 0.00028268909435223347, "epoch": 0.12, "step": 3210}
+{"loss": 0.7473, "learning_rate": 0.0002826333964576139, "epoch": 0.12, "step": 3220}
+{"loss": 0.7208, "learning_rate": 0.0002825776985629943, "epoch": 0.12, "step": 3230}
+{"loss": 0.8586, "learning_rate": 0.0002825220006683747, "epoch": 0.12, "step": 3240}
+{"loss": 0.6439, "learning_rate": 0.0002824663027737551, "epoch": 0.12, "step": 3250}
+{"loss": 0.7187, "learning_rate": 0.0002824106048791356, "epoch": 0.12, "step": 3260}
+{"loss": 0.6736, "learning_rate": 0.000282354906984516, "epoch": 0.12, "step": 3270}
+{"loss": 0.812, "learning_rate": 0.0002822992090898964, "epoch": 0.12, "step": 3280}
+{"loss": 0.7541, "learning_rate": 0.0002822435111952768, "epoch": 0.12, "step": 3290}
+{"loss": 0.7432, "learning_rate": 0.0002821878133006572, "epoch": 0.12, "step": 3300}
+{"loss": 0.7788, "learning_rate": 0.00028213211540603763, "epoch": 0.12, "step": 3310}
+{"loss": 0.8141, "learning_rate": 0.00028207641751141804, "epoch": 0.12, "step": 3320}
+{"loss": 0.7325, "learning_rate": 0.00028202071961679846, "epoch": 0.12, "step": 3330}
+{"loss": 0.7932, "learning_rate": 0.00028196502172217887, "epoch": 0.12, "step": 3340}
+{"loss": 0.8532, "learning_rate": 0.0002819093238275593, "epoch": 0.12, "step": 3350}
+{"loss": 0.7628, "learning_rate": 0.0002818536259329397, "epoch": 0.12, "step": 3360}
+{"loss": 0.7139, "learning_rate": 0.0002817979280383201, "epoch": 0.12, "step": 3370}
+{"loss": 0.7749, "learning_rate": 0.0002817422301437005, "epoch": 0.13, "step": 3380}
+{"loss": 0.6444, "learning_rate": 0.000281686532249081, "epoch": 0.13, "step": 3390}
+{"loss": 0.6748, "learning_rate": 0.0002816308343544614, "epoch": 0.13, "step": 3400}
+{"loss": 0.7428, "learning_rate": 0.0002815751364598418, "epoch": 0.13, "step": 3410}
+{"loss": 0.7912, "learning_rate": 0.0002815194385652222, "epoch": 0.13, "step": 3420}
+{"loss": 0.8587, "learning_rate": 0.0002814637406706026, "epoch": 0.13, "step": 3430}
+{"loss": 0.7408, "learning_rate": 0.00028140804277598303, "epoch": 0.13, "step": 3440}
+{"loss": 0.7386, "learning_rate": 0.00028135234488136344, "epoch": 0.13, "step": 3450}
+{"loss": 0.798, "learning_rate": 0.0002812966469867439, "epoch": 0.13, "step": 3460}
+{"loss": 0.6816, "learning_rate": 0.0002812409490921243, "epoch": 0.13, "step": 3470}
+{"loss": 0.8592, "learning_rate": 0.00028118525119750473, "epoch": 0.13, "step": 3480}
+{"loss": 0.5379, "learning_rate": 0.00028112955330288514, "epoch": 0.13, "step": 3490}
+{"loss": 0.721, "learning_rate": 0.00028107385540826555, "epoch": 0.13, "step": 3500}
+{"loss": 0.7638, "learning_rate": 0.00028101815751364596, "epoch": 0.13, "step": 3510}
+{"loss": 0.7621, "learning_rate": 0.0002809624596190264, "epoch": 0.13, "step": 3520}
+{"loss": 0.7362, "learning_rate": 0.0002809067617244068, "epoch": 0.13, "step": 3530}
+{"loss": 0.7464, "learning_rate": 0.0002808510638297872, "epoch": 0.13, "step": 3540}
+{"loss": 0.7857, "learning_rate": 0.0002807953659351676, "epoch": 0.13, "step": 3550}
+{"loss": 0.6152, "learning_rate": 0.000280739668040548, "epoch": 0.13, "step": 3560}
+{"loss": 0.6198, "learning_rate": 0.00028068397014592843, "epoch": 0.13, "step": 3570}
+{"loss": 0.7537, "learning_rate": 0.00028062827225130884, "epoch": 0.13, "step": 3580}
+{"loss": 0.8577, "learning_rate": 0.0002805725743566893, "epoch": 0.13, "step": 3590}
+{"loss": 0.649, "learning_rate": 0.0002805168764620697, "epoch": 0.13, "step": 3600}
+{"loss": 0.6553, "learning_rate": 0.00028046117856745013, "epoch": 0.13, "step": 3610}
+{"loss": 0.7438, "learning_rate": 0.00028040548067283054, "epoch": 0.13, "step": 3620}
+{"loss": 0.7611, "learning_rate": 0.00028034978277821095, "epoch": 0.13, "step": 3630}
+{"loss": 0.7373, "learning_rate": 0.00028029408488359136, "epoch": 0.13, "step": 3640}
+{"loss": 0.7569, "learning_rate": 0.00028023838698897183, "epoch": 0.14, "step": 3650}
+{"loss": 0.8564, "learning_rate": 0.00028018268909435224, "epoch": 0.14, "step": 3660}
+{"loss": 0.7556, "learning_rate": 0.00028012699119973265, "epoch": 0.14, "step": 3670}
+{"loss": 0.6129, "learning_rate": 0.00028007129330511306, "epoch": 0.14, "step": 3680}
+{"loss": 0.7645, "learning_rate": 0.0002800155954104935, "epoch": 0.14, "step": 3690}
+{"loss": 0.8212, "learning_rate": 0.0002799598975158739, "epoch": 0.14, "step": 3700}
+{"loss": 0.652, "learning_rate": 0.0002799041996212543, "epoch": 0.14, "step": 3710}
+{"loss": 0.783, "learning_rate": 0.0002798485017266347, "epoch": 0.14, "step": 3720}
+{"loss": 0.7443, "learning_rate": 0.0002797928038320151, "epoch": 0.14, "step": 3730}
+{"loss": 0.7401, "learning_rate": 0.00027973710593739553, "epoch": 0.14, "step": 3740}
+{"loss": 0.6446, "learning_rate": 0.00027968140804277594, "epoch": 0.14, "step": 3750}
+{"loss": 0.7434, "learning_rate": 0.00027962571014815635, "epoch": 0.14, "step": 3760}
+{"loss": 0.6295, "learning_rate": 0.00027957001225353676, "epoch": 0.14, "step": 3770}
+{"loss": 0.6988, "learning_rate": 0.00027951431435891723, "epoch": 0.14, "step": 3780}
+{"loss": 0.741, "learning_rate": 0.00027945861646429764, "epoch": 0.14, "step": 3790}
+{"loss": 0.8125, "learning_rate": 0.00027940291856967805, "epoch": 0.14, "step": 3800}
+{"loss": 0.8193, "learning_rate": 0.00027934722067505846, "epoch": 0.14, "step": 3810}
+{"loss": 0.734, "learning_rate": 0.00027929152278043887, "epoch": 0.14, "step": 3820}
+{"loss": 0.7431, "learning_rate": 0.0002792358248858193, "epoch": 0.14, "step": 3830}
+{"loss": 0.7123, "learning_rate": 0.0002791801269911997, "epoch": 0.14, "step": 3840}
+{"loss": 0.7364, "learning_rate": 0.00027912442909658016, "epoch": 0.14, "step": 3850}
+{"loss": 0.6552, "learning_rate": 0.00027906873120196057, "epoch": 0.14, "step": 3860}
+{"loss": 0.5693, "learning_rate": 0.000279013033307341, "epoch": 0.14, "step": 3870}
+{"loss": 0.7448, "learning_rate": 0.0002789573354127214, "epoch": 0.14, "step": 3880}
+{"loss": 0.846, "learning_rate": 0.0002789016375181018, "epoch": 0.14, "step": 3890}
+{"loss": 0.6561, "learning_rate": 0.0002788459396234822, "epoch": 0.14, "step": 3900}
+{"loss": 0.6593, "learning_rate": 0.0002787902417288626, "epoch": 0.14, "step": 3910}
+{"loss": 0.6321, "learning_rate": 0.00027873454383424304, "epoch": 0.15, "step": 3920}
+{"loss": 0.7445, "learning_rate": 0.00027867884593962345, "epoch": 0.15, "step": 3930}
+{"loss": 0.7248, "learning_rate": 0.00027862314804500386, "epoch": 0.15, "step": 3940}
+{"loss": 0.7501, "learning_rate": 0.00027856745015038427, "epoch": 0.15, "step": 3950}
+{"loss": 0.7191, "learning_rate": 0.0002785117522557647, "epoch": 0.15, "step": 3960}
+{"loss": 0.7647, "learning_rate": 0.0002784560543611451, "epoch": 0.15, "step": 3970}
+{"loss": 0.7046, "learning_rate": 0.00027840035646652556, "epoch": 0.15, "step": 3980}
+{"loss": 0.7628, "learning_rate": 0.00027834465857190597, "epoch": 0.15, "step": 3990}
+{"loss": 0.6385, "learning_rate": 0.0002782889606772864, "epoch": 0.15, "step": 4000}
+{"loss": 0.786, "learning_rate": 0.0002782332627826668, "epoch": 0.15, "step": 4010}
+{"loss": 0.7781, "learning_rate": 0.0002781775648880472, "epoch": 0.15, "step": 4020}
+{"loss": 0.6636, "learning_rate": 0.0002781218669934276, "epoch": 0.15, "step": 4030}
+{"loss": 0.8073, "learning_rate": 0.0002780661690988081, "epoch": 0.15, "step": 4040}
+{"loss": 0.6326, "learning_rate": 0.0002780104712041885, "epoch": 0.15, "step": 4050}
+{"loss": 0.5519, "learning_rate": 0.0002779547733095689, "epoch": 0.15, "step": 4060}
+{"loss": 0.7921, "learning_rate": 0.0002778990754149493, "epoch": 0.15, "step": 4070}
+{"loss": 0.7914, "learning_rate": 0.0002778433775203297, "epoch": 0.15, "step": 4080}
+{"loss": 0.744, "learning_rate": 0.00027778767962571014, "epoch": 0.15, "step": 4090}
+{"loss": 0.7316, "learning_rate": 0.00027773198173109055, "epoch": 0.15, "step": 4100}
+{"loss": 0.8279, "learning_rate": 0.00027767628383647096, "epoch": 0.15, "step": 4110}
+{"loss": 0.7324, "learning_rate": 0.00027762058594185137, "epoch": 0.15, "step": 4120}
+{"loss": 0.718, "learning_rate": 0.0002775648880472318, "epoch": 0.15, "step": 4130}
+{"loss": 0.621, "learning_rate": 0.0002775091901526122, "epoch": 0.15, "step": 4140}
+{"loss": 0.7068, "learning_rate": 0.0002774534922579926, "epoch": 0.15, "step": 4150}
+{"loss": 0.6845, "learning_rate": 0.000277397794363373, "epoch": 0.15, "step": 4160}
+{"loss": 0.785, "learning_rate": 0.0002773420964687535, "epoch": 0.15, "step": 4170}
+{"loss": 0.6549, "learning_rate": 0.0002772863985741339, "epoch": 0.15, "step": 4180}
+{"loss": 0.7149, "learning_rate": 0.0002772307006795143, "epoch": 0.16, "step": 4190}
+{"loss": 0.687, "learning_rate": 0.0002771750027848947, "epoch": 0.16, "step": 4200}
+{"loss": 0.6946, "learning_rate": 0.0002771193048902751, "epoch": 0.16, "step": 4210}
+{"loss": 0.6804, "learning_rate": 0.00027706360699565553, "epoch": 0.16, "step": 4220}
+{"loss": 0.6528, "learning_rate": 0.00027700790910103595, "epoch": 0.16, "step": 4230}
+{"loss": 0.6694, "learning_rate": 0.0002769522112064164, "epoch": 0.16, "step": 4240}
+{"loss": 0.7115, "learning_rate": 0.0002768965133117968, "epoch": 0.16, "step": 4250}
+{"loss": 0.8429, "learning_rate": 0.00027684081541717723, "epoch": 0.16, "step": 4260}
+{"loss": 0.7695, "learning_rate": 0.00027678511752255764, "epoch": 0.16, "step": 4270}
+{"loss": 0.6824, "learning_rate": 0.00027672941962793806, "epoch": 0.16, "step": 4280}
+{"loss": 0.6731, "learning_rate": 0.00027667372173331847, "epoch": 0.16, "step": 4290}
+{"loss": 0.922, "learning_rate": 0.0002766180238386989, "epoch": 0.16, "step": 4300}
+{"loss": 0.8287, "learning_rate": 0.0002765623259440793, "epoch": 0.16, "step": 4310}
+{"loss": 0.8205, "learning_rate": 0.0002765066280494597, "epoch": 0.16, "step": 4320}
+{"loss": 0.7776, "learning_rate": 0.0002764509301548401, "epoch": 0.16, "step": 4330}
+{"loss": 0.7258, "learning_rate": 0.0002763952322602205, "epoch": 0.16, "step": 4340}
+{"loss": 0.8068, "learning_rate": 0.00027633953436560093, "epoch": 0.16, "step": 4350}
+{"loss": 0.7317, "learning_rate": 0.00027628383647098134, "epoch": 0.16, "step": 4360}
+{"loss": 0.7747, "learning_rate": 0.0002762281385763618, "epoch": 0.16, "step": 4370}
+{"loss": 0.7258, "learning_rate": 0.0002761724406817422, "epoch": 0.16, "step": 4380}
+{"loss": 0.689, "learning_rate": 0.00027611674278712263, "epoch": 0.16, "step": 4390}
+{"loss": 0.7344, "learning_rate": 0.00027606104489250304, "epoch": 0.16, "step": 4400}
+{"loss": 0.6484, "learning_rate": 0.00027600534699788345, "epoch": 0.16, "step": 4410}
+{"loss": 0.6977, "learning_rate": 0.00027594964910326387, "epoch": 0.16, "step": 4420}
+{"loss": 0.6636, "learning_rate": 0.0002758939512086443, "epoch": 0.16, "step": 4430}
+{"loss": 0.75, "learning_rate": 0.00027583825331402474, "epoch": 0.16, "step": 4440}
+{"loss": 0.6786, "learning_rate": 0.00027578255541940515, "epoch": 0.16, "step": 4450}
+{"loss": 0.7433, "learning_rate": 0.00027572685752478556, "epoch": 0.17, "step": 4460}
+{"loss": 0.6137, "learning_rate": 0.000275671159630166, "epoch": 0.17, "step": 4470}
+{"loss": 0.6681, "learning_rate": 0.0002756154617355464, "epoch": 0.17, "step": 4480}
+{"loss": 0.7152, "learning_rate": 0.0002755597638409268, "epoch": 0.17, "step": 4490}
+{"loss": 0.769, "learning_rate": 0.0002755040659463072, "epoch": 0.17, "step": 4500}
+{"loss": 0.7291, "learning_rate": 0.0002754483680516876, "epoch": 0.17, "step": 4510}
+{"loss": 0.7176, "learning_rate": 0.00027539267015706803, "epoch": 0.17, "step": 4520}
+{"loss": 0.8482, "learning_rate": 0.0002753425420519104, "epoch": 0.17, "step": 4530}
+{"loss": 0.7211, "learning_rate": 0.00027528684415729086, "epoch": 0.17, "step": 4540}
+{"loss": 0.653, "learning_rate": 0.00027523114626267127, "epoch": 0.17, "step": 4550}
+{"loss": 0.7194, "learning_rate": 0.0002751754483680517, "epoch": 0.17, "step": 4560}
+{"loss": 0.8425, "learning_rate": 0.0002751197504734321, "epoch": 0.17, "step": 4570}
+{"loss": 0.6682, "learning_rate": 0.0002750640525788125, "epoch": 0.17, "step": 4580}
+{"loss": 0.7796, "learning_rate": 0.0002750083546841929, "epoch": 0.17, "step": 4590}
+{"loss": 0.8755, "learning_rate": 0.0002749526567895733, "epoch": 0.17, "step": 4600}
+{"loss": 0.6759, "learning_rate": 0.00027489695889495373, "epoch": 0.17, "step": 4610}
+{"loss": 0.8197, "learning_rate": 0.00027484126100033414, "epoch": 0.17, "step": 4620}
+{"loss": 0.7042, "learning_rate": 0.00027478556310571456, "epoch": 0.17, "step": 4630}
+{"loss": 0.765, "learning_rate": 0.00027472986521109497, "epoch": 0.17, "step": 4640}
+{"loss": 0.8929, "learning_rate": 0.0002746741673164754, "epoch": 0.17, "step": 4650}
+{"loss": 0.614, "learning_rate": 0.0002746184694218558, "epoch": 0.17, "step": 4660}
+{"loss": 0.7963, "learning_rate": 0.00027456277152723625, "epoch": 0.17, "step": 4670}
+{"loss": 0.6607, "learning_rate": 0.00027450707363261667, "epoch": 0.17, "step": 4680}
+{"loss": 0.7462, "learning_rate": 0.0002744513757379971, "epoch": 0.17, "step": 4690}
+{"loss": 0.8325, "learning_rate": 0.0002743956778433775, "epoch": 0.17, "step": 4700}
+{"loss": 0.765, "learning_rate": 0.0002743399799487579, "epoch": 0.17, "step": 4710}
+{"loss": 0.7377, "learning_rate": 0.0002742842820541383, "epoch": 0.17, "step": 4720}
+{"loss": 0.8078, "learning_rate": 0.0002742285841595188, "epoch": 0.18, "step": 4730}
+{"loss": 0.7291, "learning_rate": 0.0002741728862648992, "epoch": 0.18, "step": 4740}
+{"loss": 0.6825, "learning_rate": 0.0002741171883702796, "epoch": 0.18, "step": 4750}
+{"loss": 0.5591, "learning_rate": 0.00027406149047566, "epoch": 0.18, "step": 4760}
+{"loss": 0.6499, "learning_rate": 0.0002740057925810404, "epoch": 0.18, "step": 4770}
+{"loss": 0.74, "learning_rate": 0.00027395009468642083, "epoch": 0.18, "step": 4780}
+{"loss": 0.66, "learning_rate": 0.00027389439679180124, "epoch": 0.18, "step": 4790}
+{"loss": 0.8379, "learning_rate": 0.00027383869889718165, "epoch": 0.18, "step": 4800}
+{"loss": 0.7177, "learning_rate": 0.00027378300100256206, "epoch": 0.18, "step": 4810}
+{"loss": 0.5488, "learning_rate": 0.0002737273031079425, "epoch": 0.18, "step": 4820}
+{"loss": 0.7356, "learning_rate": 0.0002736716052133229, "epoch": 0.18, "step": 4830}
+{"loss": 0.7999, "learning_rate": 0.0002736159073187033, "epoch": 0.18, "step": 4840}
+{"loss": 0.8105, "learning_rate": 0.0002735602094240837, "epoch": 0.18, "step": 4850}
+{"loss": 0.6913, "learning_rate": 0.0002735045115294642, "epoch": 0.18, "step": 4860}
+{"loss": 0.6566, "learning_rate": 0.0002734488136348446, "epoch": 0.18, "step": 4870}
+{"loss": 0.7586, "learning_rate": 0.000273393115740225, "epoch": 0.18, "step": 4880}
+{"loss": 0.6784, "learning_rate": 0.0002733374178456054, "epoch": 0.18, "step": 4890}
+{"loss": 0.7357, "learning_rate": 0.0002732817199509858, "epoch": 0.18, "step": 4900}
+{"loss": 0.9303, "learning_rate": 0.00027322602205636623, "epoch": 0.18, "step": 4910}
+{"loss": 0.737, "learning_rate": 0.0002731703241617467, "epoch": 0.18, "step": 4920}
+{"loss": 0.7615, "learning_rate": 0.0002731146262671271, "epoch": 0.18, "step": 4930}
+{"loss": 0.7686, "learning_rate": 0.0002730589283725075, "epoch": 0.18, "step": 4940}
+{"loss": 0.6899, "learning_rate": 0.00027300323047788793, "epoch": 0.18, "step": 4950}
+{"loss": 0.6733, "learning_rate": 0.00027294753258326834, "epoch": 0.18, "step": 4960}
+{"loss": 0.7312, "learning_rate": 0.00027289183468864875, "epoch": 0.18, "step": 4970}
+{"loss": 0.5332, "learning_rate": 0.00027283613679402916, "epoch": 0.18, "step": 4980}
+{"loss": 0.764, "learning_rate": 0.0002727804388994096, "epoch": 0.18, "step": 4990}
+{"loss": 0.8746, "learning_rate": 0.00027272474100479, "epoch": 0.19, "step": 5000}
+{"loss": 0.7566, "learning_rate": 0.0002726690431101704, "epoch": 0.19, "step": 5010}
+{"loss": 0.665, "learning_rate": 0.0002726133452155508, "epoch": 0.19, "step": 5020}
+{"loss": 0.6802, "learning_rate": 0.0002725576473209312, "epoch": 0.19, "step": 5030}
+{"loss": 0.7794, "learning_rate": 0.00027250194942631163, "epoch": 0.19, "step": 5040}
+{"loss": 0.7101, "learning_rate": 0.0002724462515316921, "epoch": 0.19, "step": 5050}
+{"loss": 0.8295, "learning_rate": 0.0002723905536370725, "epoch": 0.19, "step": 5060}
+{"loss": 0.6303, "learning_rate": 0.0002723348557424529, "epoch": 0.19, "step": 5070}
+{"loss": 0.7234, "learning_rate": 0.00027227915784783333, "epoch": 0.19, "step": 5080}
+{"loss": 0.6497, "learning_rate": 0.00027222345995321374, "epoch": 0.19, "step": 5090}
+{"loss": 0.6319, "learning_rate": 0.00027216776205859415, "epoch": 0.19, "step": 5100}
+{"loss": 0.8571, "learning_rate": 0.00027211206416397456, "epoch": 0.19, "step": 5110}
+{"loss": 0.7513, "learning_rate": 0.000272056366269355, "epoch": 0.19, "step": 5120}
+{"loss": 0.6934, "learning_rate": 0.00027200066837473544, "epoch": 0.19, "step": 5130}
+{"loss": 0.739, "learning_rate": 0.00027194497048011585, "epoch": 0.19, "step": 5140}
+{"loss": 0.7325, "learning_rate": 0.00027188927258549626, "epoch": 0.19, "step": 5150}
+{"loss": 0.6974, "learning_rate": 0.00027183357469087667, "epoch": 0.19, "step": 5160}
+{"loss": 0.6632, "learning_rate": 0.0002717778767962571, "epoch": 0.19, "step": 5170}
+{"loss": 0.6815, "learning_rate": 0.0002717221789016375, "epoch": 0.19, "step": 5180}
+{"loss": 0.6149, "learning_rate": 0.0002716664810070179, "epoch": 0.19, "step": 5190}
+{"loss": 0.9114, "learning_rate": 0.0002716107831123983, "epoch": 0.19, "step": 5200}
+{"loss": 0.6524, "learning_rate": 0.00027155508521777873, "epoch": 0.19, "step": 5210}
+{"loss": 0.6619, "learning_rate": 0.00027149938732315914, "epoch": 0.19, "step": 5220}
+{"loss": 0.6642, "learning_rate": 0.00027144368942853955, "epoch": 0.19, "step": 5230}
+{"loss": 0.6233, "learning_rate": 0.00027138799153391996, "epoch": 0.19, "step": 5240}
+{"loss": 0.6848, "learning_rate": 0.0002713322936393004, "epoch": 0.19, "step": 5250}
+{"loss": 0.7088, "learning_rate": 0.00027127659574468084, "epoch": 0.19, "step": 5260}
+{"loss": 0.6855, "learning_rate": 0.00027122089785006125, "epoch": 0.2, "step": 5270}
+{"loss": 0.7053, "learning_rate": 0.00027116519995544166, "epoch": 0.2, "step": 5280}
+{"loss": 0.5642, "learning_rate": 0.00027110950206082207, "epoch": 0.2, "step": 5290}
+{"loss": 0.6653, "learning_rate": 0.0002710538041662025, "epoch": 0.2, "step": 5300}
+{"loss": 0.7754, "learning_rate": 0.00027099810627158295, "epoch": 0.2, "step": 5310}
+{"loss": 0.9231, "learning_rate": 0.00027094240837696336, "epoch": 0.2, "step": 5320}
+{"loss": 0.6057, "learning_rate": 0.00027088671048234377, "epoch": 0.2, "step": 5330}
+{"loss": 0.7264, "learning_rate": 0.0002708310125877242, "epoch": 0.2, "step": 5340}
+{"loss": 0.7683, "learning_rate": 0.0002707753146931046, "epoch": 0.2, "step": 5350}
+{"loss": 0.8335, "learning_rate": 0.000270719616798485, "epoch": 0.2, "step": 5360}
+{"loss": 0.7295, "learning_rate": 0.0002706639189038654, "epoch": 0.2, "step": 5370}
+{"loss": 0.6963, "learning_rate": 0.0002706082210092458, "epoch": 0.2, "step": 5380}
+{"loss": 0.6913, "learning_rate": 0.00027055252311462624, "epoch": 0.2, "step": 5390}
+{"loss": 0.7111, "learning_rate": 0.00027049682522000665, "epoch": 0.2, "step": 5400}
+{"loss": 0.7202, "learning_rate": 0.00027044112732538706, "epoch": 0.2, "step": 5410}
+{"loss": 0.7426, "learning_rate": 0.00027038542943076747, "epoch": 0.2, "step": 5420}
+{"loss": 0.733, "learning_rate": 0.0002703297315361479, "epoch": 0.2, "step": 5430}
+{"loss": 0.8221, "learning_rate": 0.00027027403364152835, "epoch": 0.2, "step": 5440}
+{"loss": 0.6752, "learning_rate": 0.00027021833574690876, "epoch": 0.2, "step": 5450}
+{"loss": 0.7539, "learning_rate": 0.00027016263785228917, "epoch": 0.2, "step": 5460}
+{"loss": 0.8046, "learning_rate": 0.0002701069399576696, "epoch": 0.2, "step": 5470}
+{"loss": 0.6616, "learning_rate": 0.00027005124206305, "epoch": 0.2, "step": 5480}
+{"loss": 0.6685, "learning_rate": 0.0002699955441684304, "epoch": 0.2, "step": 5490}
+{"loss": 0.5998, "learning_rate": 0.0002699398462738108, "epoch": 0.2, "step": 5500}
+{"loss": 0.6017, "learning_rate": 0.0002698841483791913, "epoch": 0.2, "step": 5510}
+{"loss": 0.7894, "learning_rate": 0.0002698284504845717, "epoch": 0.2, "step": 5520}
+{"loss": 0.7175, "learning_rate": 0.0002697727525899521, "epoch": 0.2, "step": 5530}
+{"loss": 0.6036, "learning_rate": 0.0002697170546953325, "epoch": 0.21, "step": 5540}
+{"loss": 0.8645, "learning_rate": 0.0002696613568007129, "epoch": 0.21, "step": 5550}
+{"loss": 0.7887, "learning_rate": 0.00026960565890609333, "epoch": 0.21, "step": 5560}
+{"loss": 0.792, "learning_rate": 0.00026954996101147374, "epoch": 0.21, "step": 5570}
+{"loss": 0.74, "learning_rate": 0.00026949426311685416, "epoch": 0.21, "step": 5580}
+{"loss": 0.6056, "learning_rate": 0.00026943856522223457, "epoch": 0.21, "step": 5590}
+{"loss": 0.7574, "learning_rate": 0.000269382867327615, "epoch": 0.21, "step": 5600}
+{"loss": 0.7071, "learning_rate": 0.0002693271694329954, "epoch": 0.21, "step": 5610}
+{"loss": 0.6793, "learning_rate": 0.0002692714715383758, "epoch": 0.21, "step": 5620}
+{"loss": 0.5326, "learning_rate": 0.0002692157736437562, "epoch": 0.21, "step": 5630}
+{"loss": 0.7632, "learning_rate": 0.0002691600757491367, "epoch": 0.21, "step": 5640}
+{"loss": 0.6762, "learning_rate": 0.0002691043778545171, "epoch": 0.21, "step": 5650}
+{"loss": 0.7766, "learning_rate": 0.0002690486799598975, "epoch": 0.21, "step": 5660}
+{"loss": 0.7887, "learning_rate": 0.0002689929820652779, "epoch": 0.21, "step": 5670}
+{"loss": 0.6496, "learning_rate": 0.0002689372841706583, "epoch": 0.21, "step": 5680}
+{"loss": 0.7849, "learning_rate": 0.00026888158627603873, "epoch": 0.21, "step": 5690}
+{"loss": 0.7911, "learning_rate": 0.0002688258883814192, "epoch": 0.21, "step": 5700}
+{"loss": 0.8263, "learning_rate": 0.0002687701904867996, "epoch": 0.21, "step": 5710}
+{"loss": 0.7121, "learning_rate": 0.00026871449259218, "epoch": 0.21, "step": 5720}
+{"loss": 0.76, "learning_rate": 0.00026865879469756043, "epoch": 0.21, "step": 5730}
+{"loss": 0.9227, "learning_rate": 0.00026860309680294084, "epoch": 0.21, "step": 5740}
+{"loss": 0.6894, "learning_rate": 0.00026854739890832125, "epoch": 0.21, "step": 5750}
+{"loss": 0.7586, "learning_rate": 0.00026849170101370166, "epoch": 0.21, "step": 5760}
+{"loss": 0.7109, "learning_rate": 0.0002684360031190821, "epoch": 0.21, "step": 5770}
+{"loss": 0.6974, "learning_rate": 0.0002683803052244625, "epoch": 0.21, "step": 5780}
+{"loss": 0.6584, "learning_rate": 0.0002683246073298429, "epoch": 0.21, "step": 5790}
+{"loss": 0.8414, "learning_rate": 0.0002682689094352233, "epoch": 0.21, "step": 5800}
+{"loss": 0.6988, "learning_rate": 0.0002682132115406037, "epoch": 0.22, "step": 5810}
+{"loss": 0.8674, "learning_rate": 0.00026815751364598413, "epoch": 0.22, "step": 5820}
+{"loss": 0.726, "learning_rate": 0.0002681018157513646, "epoch": 0.22, "step": 5830}
+{"loss": 0.7057, "learning_rate": 0.000268046117856745, "epoch": 0.22, "step": 5840}
+{"loss": 0.6789, "learning_rate": 0.0002679904199621254, "epoch": 0.22, "step": 5850}
+{"loss": 0.6601, "learning_rate": 0.00026793472206750583, "epoch": 0.22, "step": 5860}
+{"loss": 0.812, "learning_rate": 0.00026787902417288624, "epoch": 0.22, "step": 5870}
+{"loss": 0.6259, "learning_rate": 0.00026782332627826665, "epoch": 0.22, "step": 5880}
+{"loss": 0.5619, "learning_rate": 0.00026776762838364706, "epoch": 0.22, "step": 5890}
+{"loss": 0.7351, "learning_rate": 0.00026771193048902753, "epoch": 0.22, "step": 5900}
+{"loss": 0.557, "learning_rate": 0.00026765623259440794, "epoch": 0.22, "step": 5910}
+{"loss": 0.717, "learning_rate": 0.00026760053469978835, "epoch": 0.22, "step": 5920}
+{"loss": 0.72, "learning_rate": 0.00026754483680516876, "epoch": 0.22, "step": 5930}
+{"loss": 0.752, "learning_rate": 0.0002674891389105492, "epoch": 0.22, "step": 5940}
+{"loss": 0.7719, "learning_rate": 0.0002674334410159296, "epoch": 0.22, "step": 5950}
+{"loss": 0.7094, "learning_rate": 0.00026737774312131, "epoch": 0.22, "step": 5960}
+{"loss": 0.7023, "learning_rate": 0.0002673220452266904, "epoch": 0.22, "step": 5970}
+{"loss": 0.5916, "learning_rate": 0.0002672663473320708, "epoch": 0.22, "step": 5980}
+{"loss": 0.6855, "learning_rate": 0.00026721064943745123, "epoch": 0.22, "step": 5990}
+{"loss": 0.7081, "learning_rate": 0.00026715495154283164, "epoch": 0.22, "step": 6000}
+{"loss": 0.7359, "learning_rate": 0.00026709925364821205, "epoch": 0.22, "step": 6010}
+{"loss": 0.8214, "learning_rate": 0.00026704355575359246, "epoch": 0.22, "step": 6020}
+{"loss": 0.5878, "learning_rate": 0.00026698785785897293, "epoch": 0.22, "step": 6030}
+{"loss": 0.6977, "learning_rate": 0.00026693215996435334, "epoch": 0.22, "step": 6040}
+{"loss": 0.7332, "learning_rate": 0.00026687646206973375, "epoch": 0.22, "step": 6050}
+{"loss": 0.7437, "learning_rate": 0.00026682076417511416, "epoch": 0.22, "step": 6060}
+{"loss": 0.7216, "learning_rate": 0.00026676506628049457, "epoch": 0.22, "step": 6070}
+{"loss": 0.6907, "learning_rate": 0.000266709368385875, "epoch": 0.23, "step": 6080}
+{"loss": 0.7815, "learning_rate": 0.00026665367049125545, "epoch": 0.23, "step": 6090}
+{"loss": 0.7536, "learning_rate": 0.00026659797259663586, "epoch": 0.23, "step": 6100}
+{"loss": 0.6969, "learning_rate": 0.00026654227470201627, "epoch": 0.23, "step": 6110}
+{"loss": 0.6722, "learning_rate": 0.0002664865768073967, "epoch": 0.23, "step": 6120}
+{"loss": 0.7757, "learning_rate": 0.0002664308789127771, "epoch": 0.23, "step": 6130}
+{"loss": 0.7461, "learning_rate": 0.0002663751810181575, "epoch": 0.23, "step": 6140}
+{"loss": 0.7681, "learning_rate": 0.0002663194831235379, "epoch": 0.23, "step": 6150}
+{"loss": 0.6563, "learning_rate": 0.00026626378522891833, "epoch": 0.23, "step": 6160}
+{"loss": 0.6421, "learning_rate": 0.00026620808733429874, "epoch": 0.23, "step": 6170}
+{"loss": 0.7113, "learning_rate": 0.00026615238943967915, "epoch": 0.23, "step": 6180}
+{"loss": 0.6726, "learning_rate": 0.00026609669154505956, "epoch": 0.23, "step": 6190}
+{"loss": 0.9037, "learning_rate": 0.00026604099365043997, "epoch": 0.23, "step": 6200}
+{"loss": 0.7431, "learning_rate": 0.0002659852957558204, "epoch": 0.23, "step": 6210}
+{"loss": 0.644, "learning_rate": 0.00026592959786120085, "epoch": 0.23, "step": 6220}
+{"loss": 0.7136, "learning_rate": 0.00026587389996658126, "epoch": 0.23, "step": 6230}
+{"loss": 0.5934, "learning_rate": 0.00026581820207196167, "epoch": 0.23, "step": 6240}
+{"loss": 0.7996, "learning_rate": 0.0002657625041773421, "epoch": 0.23, "step": 6250}
+{"loss": 0.6854, "learning_rate": 0.0002657068062827225, "epoch": 0.23, "step": 6260}
+{"loss": 0.704, "learning_rate": 0.0002656511083881029, "epoch": 0.23, "step": 6270}
+{"loss": 0.7104, "learning_rate": 0.0002655954104934833, "epoch": 0.23, "step": 6280}
+{"loss": 0.6173, "learning_rate": 0.0002655397125988637, "epoch": 0.23, "step": 6290}
+{"loss": 0.6304, "learning_rate": 0.00026548401470424414, "epoch": 0.23, "step": 6300}
+{"loss": 0.7566, "learning_rate": 0.00026542831680962455, "epoch": 0.23, "step": 6310}
+{"loss": 0.8521, "learning_rate": 0.00026537261891500496, "epoch": 0.23, "step": 6320}
+{"loss": 0.6862, "learning_rate": 0.0002653169210203854, "epoch": 0.23, "step": 6330}
+{"loss": 0.6939, "learning_rate": 0.00026526122312576584, "epoch": 0.23, "step": 6340}
+{"loss": 0.6616, "learning_rate": 0.00026520552523114625, "epoch": 0.24, "step": 6350}
+{"loss": 0.6732, "learning_rate": 0.00026514982733652666, "epoch": 0.24, "step": 6360}
+{"loss": 0.6309, "learning_rate": 0.00026509412944190707, "epoch": 0.24, "step": 6370}
+{"loss": 0.6804, "learning_rate": 0.0002650384315472875, "epoch": 0.24, "step": 6380}
+{"loss": 0.5404, "learning_rate": 0.0002649827336526679, "epoch": 0.24, "step": 6390}
+{"loss": 0.7586, "learning_rate": 0.0002649270357580483, "epoch": 0.24, "step": 6400}
+{"loss": 0.7994, "learning_rate": 0.0002648713378634287, "epoch": 0.24, "step": 6410}
+{"loss": 0.6768, "learning_rate": 0.0002648156399688092, "epoch": 0.24, "step": 6420}
+{"loss": 0.6574, "learning_rate": 0.0002647599420741896, "epoch": 0.24, "step": 6430}
+{"loss": 0.7092, "learning_rate": 0.00026470424417957, "epoch": 0.24, "step": 6440}
+{"loss": 0.6939, "learning_rate": 0.0002646485462849504, "epoch": 0.24, "step": 6450}
+{"loss": 0.737, "learning_rate": 0.0002645928483903308, "epoch": 0.24, "step": 6460}
+{"loss": 0.7043, "learning_rate": 0.00026453715049571124, "epoch": 0.24, "step": 6470}
+{"loss": 0.6901, "learning_rate": 0.00026448145260109165, "epoch": 0.24, "step": 6480}
+{"loss": 0.6087, "learning_rate": 0.00026442575470647206, "epoch": 0.24, "step": 6490}
+{"loss": 0.7057, "learning_rate": 0.00026437005681185247, "epoch": 0.24, "step": 6500}
+{"loss": 0.6944, "learning_rate": 0.0002643143589172329, "epoch": 0.24, "step": 6510}
+{"loss": 0.6548, "learning_rate": 0.0002642586610226133, "epoch": 0.24, "step": 6520}
+{"loss": 0.8376, "learning_rate": 0.0002642029631279937, "epoch": 0.24, "step": 6530}
+{"loss": 0.7284, "learning_rate": 0.0002641528350228361, "epoch": 0.24, "step": 6540}
+{"loss": 0.6533, "learning_rate": 0.0002640971371282165, "epoch": 0.24, "step": 6550}
+{"loss": 0.6211, "learning_rate": 0.00026404143923359694, "epoch": 0.24, "step": 6560}
+{"loss": 0.5828, "learning_rate": 0.00026398574133897735, "epoch": 0.24, "step": 6570}
+{"loss": 0.6483, "learning_rate": 0.00026393004344435776, "epoch": 0.24, "step": 6580}
+{"loss": 0.6245, "learning_rate": 0.0002638743455497382, "epoch": 0.24, "step": 6590}
+{"loss": 0.6072, "learning_rate": 0.00026381864765511864, "epoch": 0.24, "step": 6600}
+{"loss": 0.7017, "learning_rate": 0.00026376294976049905, "epoch": 0.24, "step": 6610}
+{"loss": 0.7893, "learning_rate": 0.00026370725186587946, "epoch": 0.25, "step": 6620}
+{"loss": 0.6475, "learning_rate": 0.00026365155397125987, "epoch": 0.25, "step": 6630}
+{"loss": 0.6877, "learning_rate": 0.0002635958560766403, "epoch": 0.25, "step": 6640}
+{"loss": 0.7875, "learning_rate": 0.0002635401581820207, "epoch": 0.25, "step": 6650}
+{"loss": 0.6411, "learning_rate": 0.0002634844602874011, "epoch": 0.25, "step": 6660}
+{"loss": 0.673, "learning_rate": 0.0002634287623927815, "epoch": 0.25, "step": 6670}
+{"loss": 0.7801, "learning_rate": 0.0002633730644981619, "epoch": 0.25, "step": 6680}
+{"loss": 0.6334, "learning_rate": 0.00026331736660354234, "epoch": 0.25, "step": 6690}
+{"loss": 0.5618, "learning_rate": 0.00026326166870892275, "epoch": 0.25, "step": 6700}
+{"loss": 0.6909, "learning_rate": 0.00026320597081430316, "epoch": 0.25, "step": 6710}
+{"loss": 0.6804, "learning_rate": 0.0002631502729196836, "epoch": 0.25, "step": 6720}
+{"loss": 0.7555, "learning_rate": 0.00026309457502506404, "epoch": 0.25, "step": 6730}
+{"loss": 0.7498, "learning_rate": 0.00026303887713044445, "epoch": 0.25, "step": 6740}
+{"loss": 0.6513, "learning_rate": 0.00026298317923582486, "epoch": 0.25, "step": 6750}
+{"loss": 0.7818, "learning_rate": 0.00026292748134120527, "epoch": 0.25, "step": 6760}
+{"loss": 0.7264, "learning_rate": 0.0002628717834465857, "epoch": 0.25, "step": 6770}
+{"loss": 0.5854, "learning_rate": 0.00026281608555196615, "epoch": 0.25, "step": 6780}
+{"loss": 0.8884, "learning_rate": 0.00026276038765734656, "epoch": 0.25, "step": 6790}
+{"loss": 0.7449, "learning_rate": 0.00026270468976272697, "epoch": 0.25, "step": 6800}
+{"loss": 0.7061, "learning_rate": 0.0002626489918681074, "epoch": 0.25, "step": 6810}
+{"loss": 0.6957, "learning_rate": 0.0002625932939734878, "epoch": 0.25, "step": 6820}
+{"loss": 0.6842, "learning_rate": 0.0002625375960788682, "epoch": 0.25, "step": 6830}
+{"loss": 0.5894, "learning_rate": 0.0002624818981842486, "epoch": 0.25, "step": 6840}
+{"loss": 0.7376, "learning_rate": 0.000262426200289629, "epoch": 0.25, "step": 6850}
+{"loss": 0.6895, "learning_rate": 0.00026237050239500943, "epoch": 0.25, "step": 6860}
+{"loss": 0.7339, "learning_rate": 0.00026231480450038985, "epoch": 0.25, "step": 6870}
+{"loss": 0.6097, "learning_rate": 0.00026225910660577026, "epoch": 0.25, "step": 6880}
+{"loss": 0.6427, "learning_rate": 0.00026220340871115067, "epoch": 0.26, "step": 6890}
+{"loss": 0.7468, "learning_rate": 0.0002621477108165311, "epoch": 0.26, "step": 6900}
+{"loss": 0.5823, "learning_rate": 0.00026209201292191154, "epoch": 0.26, "step": 6910}
+{"loss": 0.6749, "learning_rate": 0.00026203631502729196, "epoch": 0.26, "step": 6920}
+{"loss": 0.7436, "learning_rate": 0.00026198061713267237, "epoch": 0.26, "step": 6930}
+{"loss": 0.7514, "learning_rate": 0.0002619249192380528, "epoch": 0.26, "step": 6940}
+{"loss": 0.7576, "learning_rate": 0.0002618692213434332, "epoch": 0.26, "step": 6950}
+{"loss": 0.6901, "learning_rate": 0.0002618135234488136, "epoch": 0.26, "step": 6960}
+{"loss": 0.6756, "learning_rate": 0.00026175782555419407, "epoch": 0.26, "step": 6970}
+{"loss": 0.6949, "learning_rate": 0.0002617021276595745, "epoch": 0.26, "step": 6980}
+{"loss": 0.7156, "learning_rate": 0.0002616464297649549, "epoch": 0.26, "step": 6990}
+{"loss": 0.7414, "learning_rate": 0.0002615907318703353, "epoch": 0.26, "step": 7000}
+{"loss": 0.7717, "learning_rate": 0.0002615350339757157, "epoch": 0.26, "step": 7010}
+{"loss": 0.7553, "learning_rate": 0.0002614793360810961, "epoch": 0.26, "step": 7020}
+{"loss": 0.628, "learning_rate": 0.00026142363818647653, "epoch": 0.26, "step": 7030}
+{"loss": 0.6224, "learning_rate": 0.00026136794029185694, "epoch": 0.26, "step": 7040}
+{"loss": 0.79, "learning_rate": 0.00026131224239723735, "epoch": 0.26, "step": 7050}
+{"loss": 0.6752, "learning_rate": 0.00026125654450261777, "epoch": 0.26, "step": 7060}
+{"loss": 0.6976, "learning_rate": 0.0002612008466079982, "epoch": 0.26, "step": 7070}
+{"loss": 0.6752, "learning_rate": 0.0002611451487133786, "epoch": 0.26, "step": 7080}
+{"loss": 0.7177, "learning_rate": 0.000261089450818759, "epoch": 0.26, "step": 7090}
+{"loss": 0.5774, "learning_rate": 0.00026103375292413946, "epoch": 0.26, "step": 7100}
+{"loss": 0.7178, "learning_rate": 0.0002609780550295199, "epoch": 0.26, "step": 7110}
+{"loss": 0.7939, "learning_rate": 0.0002609223571349003, "epoch": 0.26, "step": 7120}
+{"loss": 0.7736, "learning_rate": 0.0002608666592402807, "epoch": 0.26, "step": 7130}
+{"loss": 0.6933, "learning_rate": 0.0002608109613456611, "epoch": 0.26, "step": 7140}
+{"loss": 0.6997, "learning_rate": 0.0002607552634510415, "epoch": 0.27, "step": 7150}
+{"loss": 0.6604, "learning_rate": 0.00026069956555642193, "epoch": 0.27, "step": 7160}
+{"loss": 0.7637, "learning_rate": 0.0002606438676618024, "epoch": 0.27, "step": 7170}
+{"loss": 0.6724, "learning_rate": 0.0002605881697671828, "epoch": 0.27, "step": 7180}
+{"loss": 0.7258, "learning_rate": 0.0002605324718725632, "epoch": 0.27, "step": 7190}
+{"loss": 0.7205, "learning_rate": 0.00026047677397794363, "epoch": 0.27, "step": 7200}
+{"loss": 0.7266, "learning_rate": 0.00026042107608332404, "epoch": 0.27, "step": 7210}
+{"loss": 0.842, "learning_rate": 0.00026036537818870445, "epoch": 0.27, "step": 7220}
+{"loss": 0.7843, "learning_rate": 0.00026030968029408486, "epoch": 0.27, "step": 7230}
+{"loss": 0.6904, "learning_rate": 0.0002602539823994653, "epoch": 0.27, "step": 7240}
+{"loss": 0.6767, "learning_rate": 0.0002601982845048457, "epoch": 0.27, "step": 7250}
+{"loss": 0.7035, "learning_rate": 0.0002601425866102261, "epoch": 0.27, "step": 7260}
+{"loss": 0.7647, "learning_rate": 0.0002600868887156065, "epoch": 0.27, "step": 7270}
+{"loss": 0.6093, "learning_rate": 0.0002600311908209869, "epoch": 0.27, "step": 7280}
+{"loss": 0.6147, "learning_rate": 0.00025997549292636733, "epoch": 0.27, "step": 7290}
+{"loss": 0.7109, "learning_rate": 0.0002599197950317478, "epoch": 0.27, "step": 7300}
+{"loss": 0.7692, "learning_rate": 0.0002598640971371282, "epoch": 0.27, "step": 7310}
+{"loss": 0.654, "learning_rate": 0.0002598083992425086, "epoch": 0.27, "step": 7320}
+{"loss": 0.7219, "learning_rate": 0.00025975270134788903, "epoch": 0.27, "step": 7330}
+{"loss": 0.7073, "learning_rate": 0.00025969700345326944, "epoch": 0.27, "step": 7340}
+{"loss": 0.5878, "learning_rate": 0.00025964130555864985, "epoch": 0.27, "step": 7350}
+{"loss": 0.6496, "learning_rate": 0.0002595856076640303, "epoch": 0.27, "step": 7360}
+{"loss": 0.7443, "learning_rate": 0.00025952990976941073, "epoch": 0.27, "step": 7370}
+{"loss": 0.7504, "learning_rate": 0.00025947421187479114, "epoch": 0.27, "step": 7380}
+{"loss": 0.7563, "learning_rate": 0.00025941851398017155, "epoch": 0.27, "step": 7390}
+{"loss": 0.71, "learning_rate": 0.00025936281608555196, "epoch": 0.27, "step": 7400}
+{"loss": 0.7584, "learning_rate": 0.00025930711819093237, "epoch": 0.27, "step": 7410}
+{"loss": 0.6944, "learning_rate": 0.0002592514202963128, "epoch": 0.28, "step": 7420}
+{"loss": 0.7867, "learning_rate": 0.0002591957224016932, "epoch": 0.28, "step": 7430}
+{"loss": 0.6939, "learning_rate": 0.0002591400245070736, "epoch": 0.28, "step": 7440}
+{"loss": 0.7867, "learning_rate": 0.000259084326612454, "epoch": 0.28, "step": 7450}
+{"loss": 0.8446, "learning_rate": 0.00025902862871783443, "epoch": 0.28, "step": 7460}
+{"loss": 0.782, "learning_rate": 0.00025897293082321484, "epoch": 0.28, "step": 7470}
+{"loss": 0.818, "learning_rate": 0.00025891723292859525, "epoch": 0.28, "step": 7480}
+{"loss": 0.8101, "learning_rate": 0.0002588615350339757, "epoch": 0.28, "step": 7490}
+{"loss": 0.6433, "learning_rate": 0.0002588058371393561, "epoch": 0.28, "step": 7500}
+{"loss": 0.8316, "learning_rate": 0.00025875013924473654, "epoch": 0.28, "step": 7510}
+{"loss": 0.8513, "learning_rate": 0.00025869444135011695, "epoch": 0.28, "step": 7520}
+{"loss": 0.6972, "learning_rate": 0.00025863874345549736, "epoch": 0.28, "step": 7530}
+{"loss": 0.8297, "learning_rate": 0.00025858304556087777, "epoch": 0.28, "step": 7540}
+{"loss": 0.6686, "learning_rate": 0.0002585273476662582, "epoch": 0.28, "step": 7550}
+{"loss": 0.6718, "learning_rate": 0.00025847164977163865, "epoch": 0.28, "step": 7560}
+{"loss": 0.7151, "learning_rate": 0.00025841595187701906, "epoch": 0.28, "step": 7570}
+{"loss": 0.6683, "learning_rate": 0.00025836025398239947, "epoch": 0.28, "step": 7580}
+{"loss": 0.8528, "learning_rate": 0.0002583045560877799, "epoch": 0.28, "step": 7590}
+{"loss": 0.8118, "learning_rate": 0.0002582488581931603, "epoch": 0.28, "step": 7600}
+{"loss": 0.7875, "learning_rate": 0.0002581931602985407, "epoch": 0.28, "step": 7610}
+{"loss": 0.6881, "learning_rate": 0.0002581374624039211, "epoch": 0.28, "step": 7620}
+{"loss": 0.6904, "learning_rate": 0.0002580817645093015, "epoch": 0.28, "step": 7630}
+{"loss": 0.7501, "learning_rate": 0.00025802606661468194, "epoch": 0.28, "step": 7640}
+{"loss": 0.5125, "learning_rate": 0.00025797036872006235, "epoch": 0.28, "step": 7650}
+{"loss": 0.6751, "learning_rate": 0.00025791467082544276, "epoch": 0.28, "step": 7660}
+{"loss": 0.6571, "learning_rate": 0.00025785897293082317, "epoch": 0.28, "step": 7670}
+{"loss": 0.6458, "learning_rate": 0.0002578032750362036, "epoch": 0.28, "step": 7680}
+{"loss": 0.5871, "learning_rate": 0.00025774757714158405, "epoch": 0.29, "step": 7690}
+{"loss": 0.6758, "learning_rate": 0.00025769187924696446, "epoch": 0.29, "step": 7700}
+{"loss": 0.8249, "learning_rate": 0.00025763618135234487, "epoch": 0.29, "step": 7710}
+{"loss": 0.7329, "learning_rate": 0.0002575804834577253, "epoch": 0.29, "step": 7720}
+{"loss": 0.9108, "learning_rate": 0.0002575247855631057, "epoch": 0.29, "step": 7730}
+{"loss": 0.7389, "learning_rate": 0.0002574690876684861, "epoch": 0.29, "step": 7740}
+{"loss": 0.7714, "learning_rate": 0.00025741338977386657, "epoch": 0.29, "step": 7750}
+{"loss": 0.7028, "learning_rate": 0.000257357691879247, "epoch": 0.29, "step": 7760}
+{"loss": 0.7169, "learning_rate": 0.0002573019939846274, "epoch": 0.29, "step": 7770}
+{"loss": 0.6585, "learning_rate": 0.0002572462960900078, "epoch": 0.29, "step": 7780}
+{"loss": 0.6287, "learning_rate": 0.0002571905981953882, "epoch": 0.29, "step": 7790}
+{"loss": 0.6507, "learning_rate": 0.0002571349003007686, "epoch": 0.29, "step": 7800}
+{"loss": 0.7753, "learning_rate": 0.00025707920240614903, "epoch": 0.29, "step": 7810}
+{"loss": 0.7194, "learning_rate": 0.00025702350451152945, "epoch": 0.29, "step": 7820}
+{"loss": 0.568, "learning_rate": 0.00025696780661690986, "epoch": 0.29, "step": 7830}
+{"loss": 0.6429, "learning_rate": 0.00025691210872229027, "epoch": 0.29, "step": 7840}
+{"loss": 0.7006, "learning_rate": 0.0002568564108276707, "epoch": 0.29, "step": 7850}
+{"loss": 0.6556, "learning_rate": 0.0002568007129330511, "epoch": 0.29, "step": 7860}
+{"loss": 0.7786, "learning_rate": 0.0002567450150384315, "epoch": 0.29, "step": 7870}
+{"loss": 0.6277, "learning_rate": 0.00025668931714381197, "epoch": 0.29, "step": 7880}
+{"loss": 0.6289, "learning_rate": 0.0002566336192491924, "epoch": 0.29, "step": 7890}
+{"loss": 0.6624, "learning_rate": 0.0002565779213545728, "epoch": 0.29, "step": 7900}
+{"loss": 0.6011, "learning_rate": 0.0002565222234599532, "epoch": 0.29, "step": 7910}
+{"loss": 0.6754, "learning_rate": 0.0002564665255653336, "epoch": 0.29, "step": 7920}
+{"loss": 0.6884, "learning_rate": 0.000256410827670714, "epoch": 0.29, "step": 7930}
+{"loss": 0.5884, "learning_rate": 0.00025635512977609443, "epoch": 0.29, "step": 7940}
+{"loss": 0.6332, "learning_rate": 0.00025629943188147484, "epoch": 0.29, "step": 7950}
+{"loss": 0.7384, "learning_rate": 0.00025624373398685526, "epoch": 0.3, "step": 7960}
+{"loss": 0.8223, "learning_rate": 0.00025618803609223567, "epoch": 0.3, "step": 7970}
+{"loss": 0.6917, "learning_rate": 0.0002561323381976161, "epoch": 0.3, "step": 7980}
+{"loss": 0.6033, "learning_rate": 0.00025607664030299654, "epoch": 0.3, "step": 7990}
+{"loss": 0.644, "learning_rate": 0.00025602094240837695, "epoch": 0.3, "step": 8000}
+{"loss": 0.8084, "learning_rate": 0.00025596524451375737, "epoch": 0.3, "step": 8010}
+{"loss": 0.7939, "learning_rate": 0.0002559095466191378, "epoch": 0.3, "step": 8020}
+{"loss": 0.7747, "learning_rate": 0.00025585941851398014, "epoch": 0.3, "step": 8030}
+{"loss": 0.6367, "learning_rate": 0.00025580372061936055, "epoch": 0.3, "step": 8040}
+{"loss": 0.7061, "learning_rate": 0.000255748022724741, "epoch": 0.3, "step": 8050}
+{"loss": 0.6948, "learning_rate": 0.0002556923248301214, "epoch": 0.3, "step": 8060}
+{"loss": 0.706, "learning_rate": 0.00025563662693550183, "epoch": 0.3, "step": 8070}
+{"loss": 0.6919, "learning_rate": 0.00025558092904088225, "epoch": 0.3, "step": 8080}
+{"loss": 0.614, "learning_rate": 0.00025552523114626266, "epoch": 0.3, "step": 8090}
+{"loss": 0.7104, "learning_rate": 0.00025546953325164307, "epoch": 0.3, "step": 8100}
+{"loss": 0.6637, "learning_rate": 0.0002554138353570235, "epoch": 0.3, "step": 8110}
+{"loss": 0.7115, "learning_rate": 0.0002553581374624039, "epoch": 0.3, "step": 8120}
+{"loss": 0.7232, "learning_rate": 0.0002553024395677843, "epoch": 0.3, "step": 8130}
+{"loss": 0.7147, "learning_rate": 0.0002552467416731647, "epoch": 0.3, "step": 8140}
+{"loss": 0.6032, "learning_rate": 0.0002551910437785451, "epoch": 0.3, "step": 8150}
+{"loss": 0.6833, "learning_rate": 0.00025513534588392553, "epoch": 0.3, "step": 8160}
+{"loss": 0.783, "learning_rate": 0.00025507964798930595, "epoch": 0.3, "step": 8170}
+{"loss": 0.6464, "learning_rate": 0.0002550239500946864, "epoch": 0.3, "step": 8180}
+{"loss": 0.6631, "learning_rate": 0.0002549682522000668, "epoch": 0.3, "step": 8190}
+{"loss": 0.6535, "learning_rate": 0.00025491255430544723, "epoch": 0.3, "step": 8200}
+{"loss": 0.7087, "learning_rate": 0.00025485685641082764, "epoch": 0.3, "step": 8210}
+{"loss": 0.6446, "learning_rate": 0.00025480115851620806, "epoch": 0.3, "step": 8220}
+{"loss": 0.6835, "learning_rate": 0.00025474546062158847, "epoch": 0.31, "step": 8230}
+{"loss": 0.7274, "learning_rate": 0.0002546897627269689, "epoch": 0.31, "step": 8240}
+{"loss": 0.7993, "learning_rate": 0.00025463406483234934, "epoch": 0.31, "step": 8250}
+{"loss": 0.7761, "learning_rate": 0.00025457836693772975, "epoch": 0.31, "step": 8260}
+{"loss": 0.6708, "learning_rate": 0.00025452266904311017, "epoch": 0.31, "step": 8270}
+{"loss": 0.6827, "learning_rate": 0.0002544669711484906, "epoch": 0.31, "step": 8280}
+{"loss": 0.6779, "learning_rate": 0.000254411273253871, "epoch": 0.31, "step": 8290}
+{"loss": 0.7011, "learning_rate": 0.0002543555753592514, "epoch": 0.31, "step": 8300}
+{"loss": 0.6129, "learning_rate": 0.0002542998774646318, "epoch": 0.31, "step": 8310}
+{"loss": 0.635, "learning_rate": 0.0002542441795700122, "epoch": 0.31, "step": 8320}
+{"loss": 0.6126, "learning_rate": 0.00025418848167539263, "epoch": 0.31, "step": 8330}
+{"loss": 0.6761, "learning_rate": 0.00025413278378077304, "epoch": 0.31, "step": 8340}
+{"loss": 0.7954, "learning_rate": 0.00025407708588615345, "epoch": 0.31, "step": 8350}
+{"loss": 0.5889, "learning_rate": 0.00025402138799153387, "epoch": 0.31, "step": 8360}
+{"loss": 0.7154, "learning_rate": 0.0002539656900969143, "epoch": 0.31, "step": 8370}
+{"loss": 0.7932, "learning_rate": 0.00025390999220229474, "epoch": 0.31, "step": 8380}
+{"loss": 0.7512, "learning_rate": 0.00025385429430767515, "epoch": 0.31, "step": 8390}
+{"loss": 0.6542, "learning_rate": 0.00025379859641305556, "epoch": 0.31, "step": 8400}
+{"loss": 0.7078, "learning_rate": 0.000253742898518436, "epoch": 0.31, "step": 8410}
+{"loss": 0.7645, "learning_rate": 0.0002536872006238164, "epoch": 0.31, "step": 8420}
+{"loss": 0.7741, "learning_rate": 0.0002536315027291968, "epoch": 0.31, "step": 8430}
+{"loss": 0.791, "learning_rate": 0.00025357580483457726, "epoch": 0.31, "step": 8440}
+{"loss": 0.6548, "learning_rate": 0.0002535201069399577, "epoch": 0.31, "step": 8450}
+{"loss": 0.5454, "learning_rate": 0.0002534644090453381, "epoch": 0.31, "step": 8460}
+{"loss": 0.681, "learning_rate": 0.0002534087111507185, "epoch": 0.31, "step": 8470}
+{"loss": 0.7249, "learning_rate": 0.0002533530132560989, "epoch": 0.31, "step": 8480}
+{"loss": 0.6408, "learning_rate": 0.0002532973153614793, "epoch": 0.31, "step": 8490}
+{"loss": 0.6231, "learning_rate": 0.00025324161746685973, "epoch": 0.32, "step": 8500}
+{"loss": 0.6256, "learning_rate": 0.00025318591957224014, "epoch": 0.32, "step": 8510}
+{"loss": 0.7, "learning_rate": 0.00025313022167762055, "epoch": 0.32, "step": 8520}
+{"loss": 0.6913, "learning_rate": 0.00025307452378300096, "epoch": 0.32, "step": 8530}
+{"loss": 0.6713, "learning_rate": 0.0002530188258883814, "epoch": 0.32, "step": 8540}
+{"loss": 0.692, "learning_rate": 0.0002529631279937618, "epoch": 0.32, "step": 8550}
+{"loss": 0.6492, "learning_rate": 0.0002529074300991422, "epoch": 0.32, "step": 8560}
+{"loss": 0.7063, "learning_rate": 0.00025285173220452266, "epoch": 0.32, "step": 8570}
+{"loss": 0.6779, "learning_rate": 0.0002527960343099031, "epoch": 0.32, "step": 8580}
+{"loss": 0.6732, "learning_rate": 0.0002527403364152835, "epoch": 0.32, "step": 8590}
+{"loss": 0.6546, "learning_rate": 0.0002526846385206639, "epoch": 0.32, "step": 8600}
+{"loss": 0.6366, "learning_rate": 0.0002526289406260443, "epoch": 0.32, "step": 8610}
+{"loss": 0.6969, "learning_rate": 0.0002525732427314247, "epoch": 0.32, "step": 8620}
+{"loss": 0.7025, "learning_rate": 0.00025251754483680513, "epoch": 0.32, "step": 8630}
+{"loss": 0.7001, "learning_rate": 0.0002524618469421856, "epoch": 0.32, "step": 8640}
+{"loss": 0.6574, "learning_rate": 0.000252406149047566, "epoch": 0.32, "step": 8650}
+{"loss": 0.7517, "learning_rate": 0.0002523504511529464, "epoch": 0.32, "step": 8660}
+{"loss": 0.6742, "learning_rate": 0.00025229475325832683, "epoch": 0.32, "step": 8670}
+{"loss": 0.7066, "learning_rate": 0.00025223905536370724, "epoch": 0.32, "step": 8680}
+{"loss": 0.7134, "learning_rate": 0.00025218335746908765, "epoch": 0.32, "step": 8690}
+{"loss": 0.5786, "learning_rate": 0.00025212765957446806, "epoch": 0.32, "step": 8700}
+{"loss": 0.702, "learning_rate": 0.00025207196167984847, "epoch": 0.32, "step": 8710}
+{"loss": 0.6488, "learning_rate": 0.0002520162637852289, "epoch": 0.32, "step": 8720}
+{"loss": 0.7283, "learning_rate": 0.0002519605658906093, "epoch": 0.32, "step": 8730}
+{"loss": 0.7225, "learning_rate": 0.0002519048679959897, "epoch": 0.32, "step": 8740}
+{"loss": 0.6083, "learning_rate": 0.0002518491701013701, "epoch": 0.32, "step": 8750}
+{"loss": 0.6344, "learning_rate": 0.00025179347220675053, "epoch": 0.32, "step": 8760}
+{"loss": 0.5753, "learning_rate": 0.000251737774312131, "epoch": 0.33, "step": 8770}
+{"loss": 0.8158, "learning_rate": 0.0002516820764175114, "epoch": 0.33, "step": 8780}
+{"loss": 0.7025, "learning_rate": 0.0002516263785228918, "epoch": 0.33, "step": 8790}
+{"loss": 0.6105, "learning_rate": 0.0002515706806282722, "epoch": 0.33, "step": 8800}
+{"loss": 0.7146, "learning_rate": 0.00025151498273365264, "epoch": 0.33, "step": 8810}
+{"loss": 0.7568, "learning_rate": 0.00025145928483903305, "epoch": 0.33, "step": 8820}
+{"loss": 0.7616, "learning_rate": 0.0002514035869444135, "epoch": 0.33, "step": 8830}
+{"loss": 0.7579, "learning_rate": 0.0002513478890497939, "epoch": 0.33, "step": 8840}
+{"loss": 0.6883, "learning_rate": 0.00025129219115517434, "epoch": 0.33, "step": 8850}
+{"loss": 0.7468, "learning_rate": 0.00025123649326055475, "epoch": 0.33, "step": 8860}
+{"loss": 0.654, "learning_rate": 0.00025118079536593516, "epoch": 0.33, "step": 8870}
+{"loss": 0.709, "learning_rate": 0.00025112509747131557, "epoch": 0.33, "step": 8880}
+{"loss": 0.9533, "learning_rate": 0.000251069399576696, "epoch": 0.33, "step": 8890}
+{"loss": 0.6727, "learning_rate": 0.0002510137016820764, "epoch": 0.33, "step": 8900}
+{"loss": 0.6831, "learning_rate": 0.0002509580037874568, "epoch": 0.33, "step": 8910}
+{"loss": 0.7606, "learning_rate": 0.0002509023058928372, "epoch": 0.33, "step": 8920}
+{"loss": 0.7022, "learning_rate": 0.0002508466079982176, "epoch": 0.33, "step": 8930}
+{"loss": 0.7217, "learning_rate": 0.00025079091010359804, "epoch": 0.33, "step": 8940}
+{"loss": 0.7169, "learning_rate": 0.00025073521220897845, "epoch": 0.33, "step": 8950}
+{"loss": 0.6716, "learning_rate": 0.0002506795143143589, "epoch": 0.33, "step": 8960}
+{"loss": 0.6086, "learning_rate": 0.0002506238164197393, "epoch": 0.33, "step": 8970}
+{"loss": 0.5452, "learning_rate": 0.00025056811852511974, "epoch": 0.33, "step": 8980}
+{"loss": 0.5555, "learning_rate": 0.00025051242063050015, "epoch": 0.33, "step": 8990}
+{"loss": 0.6223, "learning_rate": 0.00025045672273588056, "epoch": 0.33, "step": 9000}
+{"loss": 0.5115, "learning_rate": 0.00025040102484126097, "epoch": 0.33, "step": 9010}
+{"loss": 0.6867, "learning_rate": 0.00025034532694664143, "epoch": 0.33, "step": 9020}
+{"loss": 0.6013, "learning_rate": 0.00025028962905202185, "epoch": 0.33, "step": 9030}
+{"loss": 0.6702, "learning_rate": 0.00025023393115740226, "epoch": 0.34, "step": 9040}
+{"loss": 0.7934, "learning_rate": 0.00025017823326278267, "epoch": 0.34, "step": 9050}
+{"loss": 0.613, "learning_rate": 0.0002501225353681631, "epoch": 0.34, "step": 9060}
+{"loss": 0.7321, "learning_rate": 0.0002500668374735435, "epoch": 0.34, "step": 9070}
+{"loss": 0.6458, "learning_rate": 0.0002500111395789239, "epoch": 0.34, "step": 9080}
+{"loss": 0.6609, "learning_rate": 0.0002499554416843043, "epoch": 0.34, "step": 9090}
+{"loss": 0.5997, "learning_rate": 0.0002498997437896847, "epoch": 0.34, "step": 9100}
+{"loss": 0.8273, "learning_rate": 0.00024984404589506513, "epoch": 0.34, "step": 9110}
+{"loss": 0.6497, "learning_rate": 0.00024978834800044555, "epoch": 0.34, "step": 9120}
+{"loss": 0.6347, "learning_rate": 0.00024973265010582596, "epoch": 0.34, "step": 9130}
+{"loss": 0.7549, "learning_rate": 0.00024967695221120637, "epoch": 0.34, "step": 9140}
+{"loss": 0.7474, "learning_rate": 0.00024962125431658683, "epoch": 0.34, "step": 9150}
+{"loss": 0.6484, "learning_rate": 0.00024956555642196724, "epoch": 0.34, "step": 9160}
+{"loss": 0.774, "learning_rate": 0.00024950985852734766, "epoch": 0.34, "step": 9170}
+{"loss": 0.6239, "learning_rate": 0.00024945416063272807, "epoch": 0.34, "step": 9180}
+{"loss": 0.6637, "learning_rate": 0.0002493984627381085, "epoch": 0.34, "step": 9190}
+{"loss": 0.703, "learning_rate": 0.0002493427648434889, "epoch": 0.34, "step": 9200}
+{"loss": 0.6582, "learning_rate": 0.0002492870669488693, "epoch": 0.34, "step": 9210}
+{"loss": 0.6731, "learning_rate": 0.00024923136905424977, "epoch": 0.34, "step": 9220}
+{"loss": 0.7721, "learning_rate": 0.0002491756711596302, "epoch": 0.34, "step": 9230}
+{"loss": 0.662, "learning_rate": 0.0002491199732650106, "epoch": 0.34, "step": 9240}
+{"loss": 0.6991, "learning_rate": 0.000249064275370391, "epoch": 0.34, "step": 9250}
+{"loss": 0.7294, "learning_rate": 0.0002490085774757714, "epoch": 0.34, "step": 9260}
+{"loss": 0.666, "learning_rate": 0.0002489528795811518, "epoch": 0.34, "step": 9270}
+{"loss": 0.7909, "learning_rate": 0.00024889718168653223, "epoch": 0.34, "step": 9280}
+{"loss": 0.8497, "learning_rate": 0.00024884148379191264, "epoch": 0.34, "step": 9290}
+{"loss": 0.7551, "learning_rate": 0.00024878578589729305, "epoch": 0.34, "step": 9300}
+{"loss": 0.6879, "learning_rate": 0.00024873008800267347, "epoch": 0.35, "step": 9310}
+{"loss": 0.7592, "learning_rate": 0.0002486743901080539, "epoch": 0.35, "step": 9320}
+{"loss": 0.6669, "learning_rate": 0.0002486186922134343, "epoch": 0.35, "step": 9330}
+{"loss": 0.5728, "learning_rate": 0.0002485629943188147, "epoch": 0.35, "step": 9340}
+{"loss": 0.6912, "learning_rate": 0.00024850729642419516, "epoch": 0.35, "step": 9350}
+{"loss": 0.82, "learning_rate": 0.0002484515985295756, "epoch": 0.35, "step": 9360}
+{"loss": 0.8179, "learning_rate": 0.000248395900634956, "epoch": 0.35, "step": 9370}
+{"loss": 0.59, "learning_rate": 0.0002483402027403364, "epoch": 0.35, "step": 9380}
+{"loss": 0.6049, "learning_rate": 0.0002482845048457168, "epoch": 0.35, "step": 9390}
+{"loss": 0.6329, "learning_rate": 0.0002482288069510972, "epoch": 0.35, "step": 9400}
+{"loss": 0.7197, "learning_rate": 0.0002481731090564777, "epoch": 0.35, "step": 9410}
+{"loss": 0.6503, "learning_rate": 0.0002481174111618581, "epoch": 0.35, "step": 9420}
+{"loss": 0.7675, "learning_rate": 0.0002480617132672385, "epoch": 0.35, "step": 9430}
+{"loss": 0.6315, "learning_rate": 0.0002480060153726189, "epoch": 0.35, "step": 9440}
+{"loss": 0.613, "learning_rate": 0.00024795031747799933, "epoch": 0.35, "step": 9450}
+{"loss": 0.7848, "learning_rate": 0.00024789461958337974, "epoch": 0.35, "step": 9460}
+{"loss": 0.7219, "learning_rate": 0.00024783892168876015, "epoch": 0.35, "step": 9470}
+{"loss": 0.5723, "learning_rate": 0.00024778322379414056, "epoch": 0.35, "step": 9480}
+{"loss": 0.7077, "learning_rate": 0.000247727525899521, "epoch": 0.35, "step": 9490}
+{"loss": 0.5787, "learning_rate": 0.0002476718280049014, "epoch": 0.35, "step": 9500}
+{"loss": 0.5726, "learning_rate": 0.0002476161301102818, "epoch": 0.35, "step": 9510}
+{"loss": 0.6345, "learning_rate": 0.0002475604322156622, "epoch": 0.35, "step": 9520}
+{"loss": 0.7327, "learning_rate": 0.0002475047343210426, "epoch": 0.35, "step": 9530}
+{"loss": 0.7187, "learning_rate": 0.0002474490364264231, "epoch": 0.35, "step": 9540}
+{"loss": 0.7135, "learning_rate": 0.0002473933385318035, "epoch": 0.35, "step": 9550}
+{"loss": 0.7937, "learning_rate": 0.0002473376406371839, "epoch": 0.35, "step": 9560}
+{"loss": 0.7391, "learning_rate": 0.0002472819427425643, "epoch": 0.35, "step": 9570}
+{"loss": 0.6772, "learning_rate": 0.00024722624484794473, "epoch": 0.36, "step": 9580}
+{"loss": 0.7811, "learning_rate": 0.00024717054695332514, "epoch": 0.36, "step": 9590}
+{"loss": 0.7126, "learning_rate": 0.00024711484905870555, "epoch": 0.36, "step": 9600}
+{"loss": 0.6437, "learning_rate": 0.00024705915116408596, "epoch": 0.36, "step": 9610}
+{"loss": 0.6263, "learning_rate": 0.0002470034532694664, "epoch": 0.36, "step": 9620}
+{"loss": 0.7226, "learning_rate": 0.0002469477553748468, "epoch": 0.36, "step": 9630}
+{"loss": 0.7301, "learning_rate": 0.00024689205748022725, "epoch": 0.36, "step": 9640}
+{"loss": 0.6653, "learning_rate": 0.00024683635958560766, "epoch": 0.36, "step": 9650}
+{"loss": 0.7842, "learning_rate": 0.00024678066169098807, "epoch": 0.36, "step": 9660}
+{"loss": 0.6937, "learning_rate": 0.0002467249637963685, "epoch": 0.36, "step": 9670}
+{"loss": 0.6652, "learning_rate": 0.0002466692659017489, "epoch": 0.36, "step": 9680}
+{"loss": 0.771, "learning_rate": 0.0002466135680071293, "epoch": 0.36, "step": 9690}
+{"loss": 0.756, "learning_rate": 0.0002465578701125097, "epoch": 0.36, "step": 9700}
+{"loss": 0.7209, "learning_rate": 0.00024650217221789013, "epoch": 0.36, "step": 9710}
+{"loss": 0.7682, "learning_rate": 0.00024644647432327054, "epoch": 0.36, "step": 9720}
+{"loss": 0.7506, "learning_rate": 0.00024639077642865095, "epoch": 0.36, "step": 9730}
+{"loss": 0.5626, "learning_rate": 0.0002463350785340314, "epoch": 0.36, "step": 9740}
+{"loss": 0.8043, "learning_rate": 0.0002462793806394118, "epoch": 0.36, "step": 9750}
+{"loss": 0.6046, "learning_rate": 0.00024622368274479224, "epoch": 0.36, "step": 9760}
+{"loss": 0.8198, "learning_rate": 0.00024616798485017265, "epoch": 0.36, "step": 9770}
+{"loss": 0.6326, "learning_rate": 0.00024611228695555306, "epoch": 0.36, "step": 9780}
+{"loss": 0.7361, "learning_rate": 0.00024605658906093347, "epoch": 0.36, "step": 9790}
+{"loss": 0.7939, "learning_rate": 0.0002460008911663139, "epoch": 0.36, "step": 9800}
+{"loss": 0.6577, "learning_rate": 0.0002459451932716943, "epoch": 0.36, "step": 9810}
+{"loss": 0.6407, "learning_rate": 0.0002458894953770747, "epoch": 0.36, "step": 9820}
+{"loss": 0.7183, "learning_rate": 0.0002458337974824551, "epoch": 0.36, "step": 9830}
+{"loss": 0.7458, "learning_rate": 0.0002457780995878355, "epoch": 0.36, "step": 9840}
+{"loss": 0.6925, "learning_rate": 0.00024572240169321594, "epoch": 0.37, "step": 9850}
+{"loss": 0.7651, "learning_rate": 0.00024566670379859635, "epoch": 0.37, "step": 9860}
+{"loss": 0.662, "learning_rate": 0.0002456110059039768, "epoch": 0.37, "step": 9870}
+{"loss": 0.6219, "learning_rate": 0.0002455553080093572, "epoch": 0.37, "step": 9880}
+{"loss": 0.6743, "learning_rate": 0.00024549961011473764, "epoch": 0.37, "step": 9890}
+{"loss": 0.6198, "learning_rate": 0.00024544391222011805, "epoch": 0.37, "step": 9900}
+{"loss": 0.7147, "learning_rate": 0.00024538821432549846, "epoch": 0.37, "step": 9910}
+{"loss": 0.8246, "learning_rate": 0.00024533251643087887, "epoch": 0.37, "step": 9920}
+{"loss": 0.702, "learning_rate": 0.00024527681853625934, "epoch": 0.37, "step": 9930}
+{"loss": 0.7672, "learning_rate": 0.00024522112064163975, "epoch": 0.37, "step": 9940}
+{"loss": 0.7388, "learning_rate": 0.00024516542274702016, "epoch": 0.37, "step": 9950}
+{"loss": 0.7253, "learning_rate": 0.00024510972485240057, "epoch": 0.37, "step": 9960}
+{"loss": 0.6616, "learning_rate": 0.000245054026957781, "epoch": 0.37, "step": 9970}
+{"loss": 0.617, "learning_rate": 0.0002449983290631614, "epoch": 0.37, "step": 9980}
+{"loss": 0.7662, "learning_rate": 0.0002449426311685418, "epoch": 0.37, "step": 9990}
+{"loss": 0.8083, "learning_rate": 0.0002448869332739222, "epoch": 0.37, "step": 10000}
+{"loss": 0.7273, "learning_rate": 0.0002448312353793026, "epoch": 0.37, "step": 10010}
+{"loss": 0.61, "learning_rate": 0.00024477553748468304, "epoch": 0.37, "step": 10020}
+{"loss": 0.8243, "learning_rate": 0.00024471983959006345, "epoch": 0.37, "step": 10030}
+{"loss": 0.7192, "learning_rate": 0.00024466414169544386, "epoch": 0.37, "step": 10040}
+{"loss": 0.4855, "learning_rate": 0.00024460844380082427, "epoch": 0.37, "step": 10050}
+{"loss": 0.707, "learning_rate": 0.00024455274590620473, "epoch": 0.37, "step": 10060}
+{"loss": 0.7075, "learning_rate": 0.00024449704801158515, "epoch": 0.37, "step": 10070}
+{"loss": 0.6282, "learning_rate": 0.00024444135011696556, "epoch": 0.37, "step": 10080}
+{"loss": 0.6695, "learning_rate": 0.00024438565222234597, "epoch": 0.37, "step": 10090}
+{"loss": 0.5489, "learning_rate": 0.0002443299543277264, "epoch": 0.37, "step": 10100}
+{"loss": 0.7393, "learning_rate": 0.0002442742564331068, "epoch": 0.37, "step": 10110}
+{"loss": 0.6557, "learning_rate": 0.0002442185585384872, "epoch": 0.38, "step": 10120}
+{"loss": 0.6952, "learning_rate": 0.00024416286064386767, "epoch": 0.38, "step": 10130}
+{"loss": 0.7737, "learning_rate": 0.00024410716274924805, "epoch": 0.38, "step": 10140}
+{"loss": 0.711, "learning_rate": 0.0002440514648546285, "epoch": 0.38, "step": 10150}
+{"loss": 0.6748, "learning_rate": 0.0002439957669600089, "epoch": 0.38, "step": 10160}
+{"loss": 0.773, "learning_rate": 0.0002439400690653893, "epoch": 0.38, "step": 10170}
+{"loss": 0.8506, "learning_rate": 0.00024388437117076972, "epoch": 0.38, "step": 10180}
+{"loss": 0.6262, "learning_rate": 0.00024382867327615016, "epoch": 0.38, "step": 10190}
+{"loss": 0.7233, "learning_rate": 0.00024377297538153057, "epoch": 0.38, "step": 10200}
+{"loss": 0.6016, "learning_rate": 0.00024371727748691098, "epoch": 0.38, "step": 10210}
+{"loss": 0.6963, "learning_rate": 0.0002436615795922914, "epoch": 0.38, "step": 10220}
+{"loss": 0.6437, "learning_rate": 0.0002436058816976718, "epoch": 0.38, "step": 10230}
+{"loss": 0.645, "learning_rate": 0.00024355018380305222, "epoch": 0.38, "step": 10240}
+{"loss": 0.6856, "learning_rate": 0.00024349448590843263, "epoch": 0.38, "step": 10250}
+{"loss": 0.6308, "learning_rate": 0.00024343878801381307, "epoch": 0.38, "step": 10260}
+{"loss": 0.6025, "learning_rate": 0.00024338309011919348, "epoch": 0.38, "step": 10270}
+{"loss": 0.6491, "learning_rate": 0.0002433273922245739, "epoch": 0.38, "step": 10280}
+{"loss": 0.6953, "learning_rate": 0.0002432716943299543, "epoch": 0.38, "step": 10290}
+{"loss": 0.641, "learning_rate": 0.0002432159964353347, "epoch": 0.38, "step": 10300}
+{"loss": 0.6919, "learning_rate": 0.00024316029854071512, "epoch": 0.38, "step": 10310}
+{"loss": 0.7419, "learning_rate": 0.00024310460064609556, "epoch": 0.38, "step": 10320}
+{"loss": 0.639, "learning_rate": 0.00024304890275147597, "epoch": 0.38, "step": 10330}
+{"loss": 0.6623, "learning_rate": 0.00024299320485685638, "epoch": 0.38, "step": 10340}
+{"loss": 0.5978, "learning_rate": 0.0002429375069622368, "epoch": 0.38, "step": 10350}
+{"loss": 0.5833, "learning_rate": 0.0002428818090676172, "epoch": 0.38, "step": 10360}
+{"loss": 0.6002, "learning_rate": 0.00024282611117299762, "epoch": 0.38, "step": 10370}
+{"loss": 0.7811, "learning_rate": 0.00024277041327837803, "epoch": 0.38, "step": 10380}
+{"loss": 0.5564, "learning_rate": 0.0002427147153837585, "epoch": 0.39, "step": 10390}
+{"loss": 0.8002, "learning_rate": 0.0002426590174891389, "epoch": 0.39, "step": 10400}
+{"loss": 0.685, "learning_rate": 0.00024260331959451931, "epoch": 0.39, "step": 10410}
+{"loss": 0.766, "learning_rate": 0.00024254762169989973, "epoch": 0.39, "step": 10420}
+{"loss": 0.7783, "learning_rate": 0.00024249192380528014, "epoch": 0.39, "step": 10430}
+{"loss": 0.8144, "learning_rate": 0.00024243622591066055, "epoch": 0.39, "step": 10440}
+{"loss": 0.6686, "learning_rate": 0.00024238052801604099, "epoch": 0.39, "step": 10450}
+{"loss": 0.7232, "learning_rate": 0.0002423248301214214, "epoch": 0.39, "step": 10460}
+{"loss": 0.7746, "learning_rate": 0.0002422691322268018, "epoch": 0.39, "step": 10470}
+{"loss": 0.8329, "learning_rate": 0.00024221343433218222, "epoch": 0.39, "step": 10480}
+{"loss": 0.6367, "learning_rate": 0.00024215773643756263, "epoch": 0.39, "step": 10490}
+{"loss": 0.7414, "learning_rate": 0.00024210203854294304, "epoch": 0.39, "step": 10500}
+{"loss": 0.7459, "learning_rate": 0.00024204634064832345, "epoch": 0.39, "step": 10510}
+{"loss": 0.5564, "learning_rate": 0.0002419906427537039, "epoch": 0.39, "step": 10520}
+{"loss": 0.7468, "learning_rate": 0.0002419349448590843, "epoch": 0.39, "step": 10530}
+{"loss": 0.6507, "learning_rate": 0.0002418792469644647, "epoch": 0.39, "step": 10540}
+{"loss": 0.7085, "learning_rate": 0.00024182354906984512, "epoch": 0.39, "step": 10550}
+{"loss": 0.6424, "learning_rate": 0.00024176785117522554, "epoch": 0.39, "step": 10560}
+{"loss": 0.7305, "learning_rate": 0.00024171215328060595, "epoch": 0.39, "step": 10570}
+{"loss": 0.7546, "learning_rate": 0.0002416564553859864, "epoch": 0.39, "step": 10580}
+{"loss": 0.705, "learning_rate": 0.00024160075749136682, "epoch": 0.39, "step": 10590}
+{"loss": 0.6509, "learning_rate": 0.00024154505959674723, "epoch": 0.39, "step": 10600}
+{"loss": 0.6708, "learning_rate": 0.00024148936170212765, "epoch": 0.39, "step": 10610}
+{"loss": 0.7686, "learning_rate": 0.00024143366380750806, "epoch": 0.39, "step": 10620}
+{"loss": 0.714, "learning_rate": 0.00024137796591288847, "epoch": 0.39, "step": 10630}
+{"loss": 0.6741, "learning_rate": 0.00024132226801826888, "epoch": 0.39, "step": 10640}
+{"loss": 0.6817, "learning_rate": 0.00024126657012364932, "epoch": 0.39, "step": 10650}
+{"loss": 0.6172, "learning_rate": 0.00024121087222902973, "epoch": 0.4, "step": 10660}
+{"loss": 0.6626, "learning_rate": 0.00024115517433441014, "epoch": 0.4, "step": 10670}
+{"loss": 0.6571, "learning_rate": 0.00024109947643979055, "epoch": 0.4, "step": 10680}
+{"loss": 0.7406, "learning_rate": 0.00024104377854517096, "epoch": 0.4, "step": 10690}
+{"loss": 0.7141, "learning_rate": 0.00024098808065055137, "epoch": 0.4, "step": 10700}
+{"loss": 0.7329, "learning_rate": 0.0002409323827559318, "epoch": 0.4, "step": 10710}
+{"loss": 0.7647, "learning_rate": 0.00024087668486131222, "epoch": 0.4, "step": 10720}
+{"loss": 0.6834, "learning_rate": 0.00024082098696669263, "epoch": 0.4, "step": 10730}
+{"loss": 0.7599, "learning_rate": 0.00024076528907207304, "epoch": 0.4, "step": 10740}
+{"loss": 0.6467, "learning_rate": 0.00024070959117745346, "epoch": 0.4, "step": 10750}
+{"loss": 0.7246, "learning_rate": 0.00024065389328283387, "epoch": 0.4, "step": 10760}
+{"loss": 0.657, "learning_rate": 0.00024059819538821433, "epoch": 0.4, "step": 10770}
+{"loss": 0.7094, "learning_rate": 0.00024054249749359474, "epoch": 0.4, "step": 10780}
+{"loss": 0.584, "learning_rate": 0.00024048679959897515, "epoch": 0.4, "step": 10790}
+{"loss": 0.8219, "learning_rate": 0.00024043110170435557, "epoch": 0.4, "step": 10800}
+{"loss": 0.6759, "learning_rate": 0.00024037540380973598, "epoch": 0.4, "step": 10810}
+{"loss": 0.8561, "learning_rate": 0.0002403197059151164, "epoch": 0.4, "step": 10820}
+{"loss": 0.724, "learning_rate": 0.0002402640080204968, "epoch": 0.4, "step": 10830}
+{"loss": 0.7839, "learning_rate": 0.00024020831012587724, "epoch": 0.4, "step": 10840}
+{"loss": 0.7722, "learning_rate": 0.00024015261223125765, "epoch": 0.4, "step": 10850}
+{"loss": 0.7929, "learning_rate": 0.00024009691433663806, "epoch": 0.4, "step": 10860}
+{"loss": 0.6527, "learning_rate": 0.00024004121644201847, "epoch": 0.4, "step": 10870}
+{"loss": 0.708, "learning_rate": 0.00023998551854739888, "epoch": 0.4, "step": 10880}
+{"loss": 0.6704, "learning_rate": 0.0002399298206527793, "epoch": 0.4, "step": 10890}
+{"loss": 0.8032, "learning_rate": 0.00023987412275815973, "epoch": 0.4, "step": 10900}
+{"loss": 0.6628, "learning_rate": 0.00023981842486354014, "epoch": 0.4, "step": 10910}
+{"loss": 0.7161, "learning_rate": 0.00023976272696892055, "epoch": 0.4, "step": 10920}
+{"loss": 0.7405, "learning_rate": 0.00023970702907430096, "epoch": 0.41, "step": 10930}
+{"loss": 0.6299, "learning_rate": 0.00023965133117968138, "epoch": 0.41, "step": 10940}
+{"loss": 0.641, "learning_rate": 0.0002395956332850618, "epoch": 0.41, "step": 10950}
+{"loss": 0.788, "learning_rate": 0.0002395399353904422, "epoch": 0.41, "step": 10960}
+{"loss": 0.7103, "learning_rate": 0.00023948423749582266, "epoch": 0.41, "step": 10970}
+{"loss": 0.7394, "learning_rate": 0.00023942853960120307, "epoch": 0.41, "step": 10980}
+{"loss": 0.6484, "learning_rate": 0.00023937284170658349, "epoch": 0.41, "step": 10990}
+{"loss": 0.6061, "learning_rate": 0.0002393171438119639, "epoch": 0.41, "step": 11000}
+{"loss": 0.7456, "learning_rate": 0.0002392614459173443, "epoch": 0.41, "step": 11010}
+{"loss": 0.6627, "learning_rate": 0.00023920574802272472, "epoch": 0.41, "step": 11020}
+{"loss": 0.5394, "learning_rate": 0.00023915005012810516, "epoch": 0.41, "step": 11030}
+{"loss": 0.7376, "learning_rate": 0.00023909435223348557, "epoch": 0.41, "step": 11040}
+{"loss": 0.841, "learning_rate": 0.00023903865433886598, "epoch": 0.41, "step": 11050}
+{"loss": 0.5351, "learning_rate": 0.0002389829564442464, "epoch": 0.41, "step": 11060}
+{"loss": 0.7919, "learning_rate": 0.0002389272585496268, "epoch": 0.41, "step": 11070}
+{"loss": 0.6437, "learning_rate": 0.0002388715606550072, "epoch": 0.41, "step": 11080}
+{"loss": 0.6065, "learning_rate": 0.00023881586276038762, "epoch": 0.41, "step": 11090}
+{"loss": 0.7801, "learning_rate": 0.00023876016486576806, "epoch": 0.41, "step": 11100}
+{"loss": 0.6902, "learning_rate": 0.00023870446697114847, "epoch": 0.41, "step": 11110}
+{"loss": 0.5621, "learning_rate": 0.00023864876907652888, "epoch": 0.41, "step": 11120}
+{"loss": 0.6552, "learning_rate": 0.0002385930711819093, "epoch": 0.41, "step": 11130}
+{"loss": 0.6235, "learning_rate": 0.0002385373732872897, "epoch": 0.41, "step": 11140}
+{"loss": 0.6562, "learning_rate": 0.00023848167539267012, "epoch": 0.41, "step": 11150}
+{"loss": 0.7234, "learning_rate": 0.00023842597749805056, "epoch": 0.41, "step": 11160}
+{"loss": 0.6843, "learning_rate": 0.00023837027960343097, "epoch": 0.41, "step": 11170}
+{"loss": 0.7269, "learning_rate": 0.0002383145817088114, "epoch": 0.41, "step": 11180}
+{"loss": 0.7298, "learning_rate": 0.00023825888381419182, "epoch": 0.41, "step": 11190}
+{"loss": 0.7556, "learning_rate": 0.00023820318591957223, "epoch": 0.42, "step": 11200}
+{"loss": 0.5612, "learning_rate": 0.00023814748802495264, "epoch": 0.42, "step": 11210}
+{"loss": 0.7553, "learning_rate": 0.00023809179013033305, "epoch": 0.42, "step": 11220}
+{"loss": 0.7956, "learning_rate": 0.0002380360922357135, "epoch": 0.42, "step": 11230}
+{"loss": 0.6301, "learning_rate": 0.0002379803943410939, "epoch": 0.42, "step": 11240}
+{"loss": 0.5737, "learning_rate": 0.0002379246964464743, "epoch": 0.42, "step": 11250}
+{"loss": 0.6406, "learning_rate": 0.00023786899855185472, "epoch": 0.42, "step": 11260}
+{"loss": 0.7076, "learning_rate": 0.00023781330065723513, "epoch": 0.42, "step": 11270}
+{"loss": 0.7455, "learning_rate": 0.00023775760276261554, "epoch": 0.42, "step": 11280}
+{"loss": 0.7002, "learning_rate": 0.00023770190486799598, "epoch": 0.42, "step": 11290}
+{"loss": 0.6286, "learning_rate": 0.0002376462069733764, "epoch": 0.42, "step": 11300}
+{"loss": 0.756, "learning_rate": 0.0002375905090787568, "epoch": 0.42, "step": 11310}
+{"loss": 0.6186, "learning_rate": 0.00023753481118413722, "epoch": 0.42, "step": 11320}
+{"loss": 0.7172, "learning_rate": 0.00023747911328951763, "epoch": 0.42, "step": 11330}
+{"loss": 0.7271, "learning_rate": 0.00023742341539489804, "epoch": 0.42, "step": 11340}
+{"loss": 0.7748, "learning_rate": 0.00023736771750027845, "epoch": 0.42, "step": 11350}
+{"loss": 0.5905, "learning_rate": 0.0002373120196056589, "epoch": 0.42, "step": 11360}
+{"loss": 0.6501, "learning_rate": 0.0002372563217110393, "epoch": 0.42, "step": 11370}
+{"loss": 0.6974, "learning_rate": 0.0002372006238164197, "epoch": 0.42, "step": 11380}
+{"loss": 0.7366, "learning_rate": 0.00023714492592180012, "epoch": 0.42, "step": 11390}
+{"loss": 0.6285, "learning_rate": 0.00023708922802718053, "epoch": 0.42, "step": 11400}
+{"loss": 0.655, "learning_rate": 0.00023703353013256094, "epoch": 0.42, "step": 11410}
+{"loss": 0.6715, "learning_rate": 0.0002369778322379414, "epoch": 0.42, "step": 11420}
+{"loss": 0.7881, "learning_rate": 0.00023692213434332182, "epoch": 0.42, "step": 11430}
+{"loss": 0.7588, "learning_rate": 0.00023686643644870223, "epoch": 0.42, "step": 11440}
+{"loss": 0.733, "learning_rate": 0.00023681073855408264, "epoch": 0.42, "step": 11450}
+{"loss": 0.7324, "learning_rate": 0.00023675504065946305, "epoch": 0.42, "step": 11460}
+{"loss": 0.7302, "learning_rate": 0.00023669934276484346, "epoch": 0.43, "step": 11470}
+{"loss": 0.6896, "learning_rate": 0.00023664364487022387, "epoch": 0.43, "step": 11480}
+{"loss": 0.7044, "learning_rate": 0.0002365879469756043, "epoch": 0.43, "step": 11490}
+{"loss": 0.6727, "learning_rate": 0.00023653224908098472, "epoch": 0.43, "step": 11500}
+{"loss": 0.593, "learning_rate": 0.00023647655118636514, "epoch": 0.43, "step": 11510}
+{"loss": 0.5973, "learning_rate": 0.00023642085329174555, "epoch": 0.43, "step": 11520}
+{"loss": 0.672, "learning_rate": 0.00023636515539712596, "epoch": 0.43, "step": 11530}
+{"loss": 0.6067, "learning_rate": 0.00023630945750250637, "epoch": 0.43, "step": 11540}
+{"loss": 0.6839, "learning_rate": 0.0002362537596078868, "epoch": 0.43, "step": 11550}
+{"loss": 0.6733, "learning_rate": 0.00023619806171326722, "epoch": 0.43, "step": 11560}
+{"loss": 0.7261, "learning_rate": 0.00023614236381864763, "epoch": 0.43, "step": 11570}
+{"loss": 0.8264, "learning_rate": 0.00023608666592402804, "epoch": 0.43, "step": 11580}
+{"loss": 0.8673, "learning_rate": 0.00023603096802940845, "epoch": 0.43, "step": 11590}
+{"loss": 0.7077, "learning_rate": 0.00023597527013478886, "epoch": 0.43, "step": 11600}
+{"loss": 0.7507, "learning_rate": 0.00023591957224016927, "epoch": 0.43, "step": 11610}
+{"loss": 0.5535, "learning_rate": 0.00023586387434554974, "epoch": 0.43, "step": 11620}
+{"loss": 0.6683, "learning_rate": 0.00023580817645093015, "epoch": 0.43, "step": 11630}
+{"loss": 0.5623, "learning_rate": 0.00023575247855631056, "epoch": 0.43, "step": 11640}
+{"loss": 0.6956, "learning_rate": 0.00023569678066169097, "epoch": 0.43, "step": 11650}
+{"loss": 0.6795, "learning_rate": 0.00023564108276707138, "epoch": 0.43, "step": 11660}
+{"loss": 0.6907, "learning_rate": 0.0002355853848724518, "epoch": 0.43, "step": 11670}
+{"loss": 0.6775, "learning_rate": 0.00023552968697783223, "epoch": 0.43, "step": 11680}
+{"loss": 0.8199, "learning_rate": 0.00023547398908321264, "epoch": 0.43, "step": 11690}
+{"loss": 0.637, "learning_rate": 0.00023541829118859306, "epoch": 0.43, "step": 11700}
+{"loss": 0.779, "learning_rate": 0.00023536259329397347, "epoch": 0.43, "step": 11710}
+{"loss": 0.5617, "learning_rate": 0.00023530689539935388, "epoch": 0.43, "step": 11720}
+{"loss": 0.562, "learning_rate": 0.0002352511975047343, "epoch": 0.43, "step": 11730}
+{"loss": 0.585, "learning_rate": 0.0002351954996101147, "epoch": 0.44, "step": 11740}
+{"loss": 0.6664, "learning_rate": 0.00023513980171549514, "epoch": 0.44, "step": 11750}
+{"loss": 0.6776, "learning_rate": 0.00023508410382087555, "epoch": 0.44, "step": 11760}
+{"loss": 0.5982, "learning_rate": 0.00023502840592625596, "epoch": 0.44, "step": 11770}
+{"loss": 0.6264, "learning_rate": 0.00023497270803163637, "epoch": 0.44, "step": 11780}
+{"loss": 0.6981, "learning_rate": 0.00023491701013701678, "epoch": 0.44, "step": 11790}
+{"loss": 0.5879, "learning_rate": 0.0002348613122423972, "epoch": 0.44, "step": 11800}
+{"loss": 0.6335, "learning_rate": 0.00023480561434777766, "epoch": 0.44, "step": 11810}
+{"loss": 0.7219, "learning_rate": 0.00023474991645315807, "epoch": 0.44, "step": 11820}
+{"loss": 0.6749, "learning_rate": 0.00023469421855853848, "epoch": 0.44, "step": 11830}
+{"loss": 0.6922, "learning_rate": 0.0002346385206639189, "epoch": 0.44, "step": 11840}
+{"loss": 0.5973, "learning_rate": 0.0002345828227692993, "epoch": 0.44, "step": 11850}
+{"loss": 0.7362, "learning_rate": 0.00023452712487467971, "epoch": 0.44, "step": 11860}
+{"loss": 0.758, "learning_rate": 0.00023447142698006013, "epoch": 0.44, "step": 11870}
+{"loss": 0.6899, "learning_rate": 0.00023441572908544056, "epoch": 0.44, "step": 11880}
+{"loss": 0.6532, "learning_rate": 0.00023436003119082098, "epoch": 0.44, "step": 11890}
+{"loss": 0.7569, "learning_rate": 0.0002343043332962014, "epoch": 0.44, "step": 11900}
+{"loss": 0.6823, "learning_rate": 0.0002342486354015818, "epoch": 0.44, "step": 11910}
+{"loss": 0.662, "learning_rate": 0.0002341929375069622, "epoch": 0.44, "step": 11920}
+{"loss": 0.6535, "learning_rate": 0.00023413723961234262, "epoch": 0.44, "step": 11930}
+{"loss": 0.7318, "learning_rate": 0.00023408154171772306, "epoch": 0.44, "step": 11940}
+{"loss": 0.702, "learning_rate": 0.00023402584382310347, "epoch": 0.44, "step": 11950}
+{"loss": 0.8139, "learning_rate": 0.00023397014592848388, "epoch": 0.44, "step": 11960}
+{"loss": 0.5758, "learning_rate": 0.0002339144480338643, "epoch": 0.44, "step": 11970}
+{"loss": 0.6429, "learning_rate": 0.0002338587501392447, "epoch": 0.44, "step": 11980}
+{"loss": 0.7322, "learning_rate": 0.00023380305224462511, "epoch": 0.44, "step": 11990}
+{"loss": 0.5917, "learning_rate": 0.00023374735435000553, "epoch": 0.44, "step": 12000}
+{"loss": 0.6779, "learning_rate": 0.000233691656455386, "epoch": 0.45, "step": 12010}
+{"loss": 0.5901, "learning_rate": 0.0002336359585607664, "epoch": 0.45, "step": 12020}
+{"loss": 0.6642, "learning_rate": 0.0002335802606661468, "epoch": 0.45, "step": 12030}
+{"loss": 0.6308, "learning_rate": 0.00023352456277152722, "epoch": 0.45, "step": 12040}
+{"loss": 0.7566, "learning_rate": 0.00023346886487690763, "epoch": 0.45, "step": 12050}
+{"loss": 0.6343, "learning_rate": 0.00023341316698228805, "epoch": 0.45, "step": 12060}
+{"loss": 0.6906, "learning_rate": 0.00023335746908766848, "epoch": 0.45, "step": 12070}
+{"loss": 0.5773, "learning_rate": 0.0002333017711930489, "epoch": 0.45, "step": 12080}
+{"loss": 0.7105, "learning_rate": 0.0002332460732984293, "epoch": 0.45, "step": 12090}
+{"loss": 0.7163, "learning_rate": 0.00023319037540380972, "epoch": 0.45, "step": 12100}
+{"loss": 0.6924, "learning_rate": 0.0002331402472986521, "epoch": 0.45, "step": 12110}
+{"loss": 0.6674, "learning_rate": 0.00023308454940403252, "epoch": 0.45, "step": 12120}
+{"loss": 0.6818, "learning_rate": 0.00023302885150941293, "epoch": 0.45, "step": 12130}
+{"loss": 0.6678, "learning_rate": 0.00023297315361479334, "epoch": 0.45, "step": 12140}
+{"loss": 0.7061, "learning_rate": 0.00023291745572017375, "epoch": 0.45, "step": 12150}
+{"loss": 0.6423, "learning_rate": 0.00023286175782555416, "epoch": 0.45, "step": 12160}
+{"loss": 0.698, "learning_rate": 0.00023280605993093457, "epoch": 0.45, "step": 12170}
+{"loss": 0.6791, "learning_rate": 0.000232750362036315, "epoch": 0.45, "step": 12180}
+{"loss": 0.6042, "learning_rate": 0.00023269466414169542, "epoch": 0.45, "step": 12190}
+{"loss": 0.6702, "learning_rate": 0.00023263896624707583, "epoch": 0.45, "step": 12200}
+{"loss": 0.6973, "learning_rate": 0.00023258326835245624, "epoch": 0.45, "step": 12210}
+{"loss": 0.6986, "learning_rate": 0.00023252757045783665, "epoch": 0.45, "step": 12220}
+{"loss": 0.6119, "learning_rate": 0.00023247187256321706, "epoch": 0.45, "step": 12230}
+{"loss": 0.7303, "learning_rate": 0.00023241617466859753, "epoch": 0.45, "step": 12240}
+{"loss": 0.8158, "learning_rate": 0.00023236047677397794, "epoch": 0.45, "step": 12250}
+{"loss": 0.625, "learning_rate": 0.00023230477887935835, "epoch": 0.45, "step": 12260}
+{"loss": 0.675, "learning_rate": 0.00023224908098473876, "epoch": 0.45, "step": 12270}
+{"loss": 0.6737, "learning_rate": 0.00023219338309011917, "epoch": 0.46, "step": 12280}
+{"loss": 0.8781, "learning_rate": 0.00023213768519549959, "epoch": 0.46, "step": 12290}
+{"loss": 0.6374, "learning_rate": 0.00023208198730088, "epoch": 0.46, "step": 12300}
+{"loss": 0.6831, "learning_rate": 0.00023202628940626044, "epoch": 0.46, "step": 12310}
+{"loss": 0.6503, "learning_rate": 0.00023197059151164085, "epoch": 0.46, "step": 12320}
+{"loss": 0.607, "learning_rate": 0.00023191489361702126, "epoch": 0.46, "step": 12330}
+{"loss": 0.6086, "learning_rate": 0.00023185919572240167, "epoch": 0.46, "step": 12340}
+{"loss": 0.736, "learning_rate": 0.00023180349782778208, "epoch": 0.46, "step": 12350}
+{"loss": 0.6346, "learning_rate": 0.0002317477999331625, "epoch": 0.46, "step": 12360}
+{"loss": 0.7117, "learning_rate": 0.00023169210203854293, "epoch": 0.46, "step": 12370}
+{"loss": 0.6165, "learning_rate": 0.00023163640414392334, "epoch": 0.46, "step": 12380}
+{"loss": 0.7097, "learning_rate": 0.00023158070624930375, "epoch": 0.46, "step": 12390}
+{"loss": 0.7594, "learning_rate": 0.00023152500835468416, "epoch": 0.46, "step": 12400}
+{"loss": 0.662, "learning_rate": 0.00023146931046006457, "epoch": 0.46, "step": 12410}
+{"loss": 0.6289, "learning_rate": 0.00023141361256544498, "epoch": 0.46, "step": 12420}
+{"loss": 0.7139, "learning_rate": 0.0002313579146708254, "epoch": 0.46, "step": 12430}
+{"loss": 0.7751, "learning_rate": 0.00023130221677620586, "epoch": 0.46, "step": 12440}
+{"loss": 0.7929, "learning_rate": 0.00023124651888158627, "epoch": 0.46, "step": 12450}
+{"loss": 0.5422, "learning_rate": 0.00023119082098696668, "epoch": 0.46, "step": 12460}
+{"loss": 0.7644, "learning_rate": 0.0002311351230923471, "epoch": 0.46, "step": 12470}
+{"loss": 0.5648, "learning_rate": 0.0002310794251977275, "epoch": 0.46, "step": 12480}
+{"loss": 0.715, "learning_rate": 0.00023102372730310792, "epoch": 0.46, "step": 12490}
+{"loss": 0.8608, "learning_rate": 0.00023096802940848836, "epoch": 0.46, "step": 12500}
+{"loss": 0.6886, "learning_rate": 0.00023091233151386877, "epoch": 0.46, "step": 12510}
+{"loss": 0.5238, "learning_rate": 0.00023085663361924918, "epoch": 0.46, "step": 12520}
+{"loss": 0.7138, "learning_rate": 0.0002308009357246296, "epoch": 0.46, "step": 12530}
+{"loss": 0.6431, "learning_rate": 0.00023074523783001, "epoch": 0.46, "step": 12540}
+{"loss": 0.8476, "learning_rate": 0.0002306895399353904, "epoch": 0.47, "step": 12550}
+{"loss": 0.6158, "learning_rate": 0.00023063384204077082, "epoch": 0.47, "step": 12560}
+{"loss": 0.678, "learning_rate": 0.00023057814414615126, "epoch": 0.47, "step": 12570}
+{"loss": 0.7354, "learning_rate": 0.00023052244625153167, "epoch": 0.47, "step": 12580}
+{"loss": 0.775, "learning_rate": 0.00023046674835691208, "epoch": 0.47, "step": 12590}
+{"loss": 0.724, "learning_rate": 0.0002304110504622925, "epoch": 0.47, "step": 12600}
+{"loss": 0.5052, "learning_rate": 0.0002303553525676729, "epoch": 0.47, "step": 12610}
+{"loss": 0.7101, "learning_rate": 0.00023029965467305332, "epoch": 0.47, "step": 12620}
+{"loss": 0.7345, "learning_rate": 0.00023024395677843378, "epoch": 0.47, "step": 12630}
+{"loss": 0.7054, "learning_rate": 0.0002301882588838142, "epoch": 0.47, "step": 12640}
+{"loss": 0.7076, "learning_rate": 0.0002301325609891946, "epoch": 0.47, "step": 12650}
+{"loss": 0.6755, "learning_rate": 0.00023007686309457501, "epoch": 0.47, "step": 12660}
+{"loss": 0.6861, "learning_rate": 0.00023002116519995543, "epoch": 0.47, "step": 12670}
+{"loss": 0.6241, "learning_rate": 0.00022996546730533584, "epoch": 0.47, "step": 12680}
+{"loss": 0.8885, "learning_rate": 0.00022990976941071628, "epoch": 0.47, "step": 12690}
+{"loss": 0.6576, "learning_rate": 0.00022985407151609669, "epoch": 0.47, "step": 12700}
+{"loss": 0.6509, "learning_rate": 0.0002297983736214771, "epoch": 0.47, "step": 12710}
+{"loss": 0.6496, "learning_rate": 0.0002297426757268575, "epoch": 0.47, "step": 12720}
+{"loss": 0.6482, "learning_rate": 0.00022968697783223792, "epoch": 0.47, "step": 12730}
+{"loss": 0.5704, "learning_rate": 0.00022963127993761833, "epoch": 0.47, "step": 12740}
+{"loss": 0.6875, "learning_rate": 0.00022957558204299874, "epoch": 0.47, "step": 12750}
+{"loss": 0.5182, "learning_rate": 0.00022951988414837918, "epoch": 0.47, "step": 12760}
+{"loss": 0.603, "learning_rate": 0.0002294641862537596, "epoch": 0.47, "step": 12770}
+{"loss": 0.6699, "learning_rate": 0.00022940848835914, "epoch": 0.47, "step": 12780}
+{"loss": 0.9298, "learning_rate": 0.00022935279046452041, "epoch": 0.47, "step": 12790}
+{"loss": 0.7229, "learning_rate": 0.00022929709256990082, "epoch": 0.47, "step": 12800}
+{"loss": 0.7195, "learning_rate": 0.00022924139467528124, "epoch": 0.47, "step": 12810}
+{"loss": 0.643, "learning_rate": 0.00022918569678066167, "epoch": 0.48, "step": 12820}
+{"loss": 0.6373, "learning_rate": 0.00022912999888604209, "epoch": 0.48, "step": 12830}
+{"loss": 0.8089, "learning_rate": 0.00022907430099142252, "epoch": 0.48, "step": 12840}
+{"loss": 0.6898, "learning_rate": 0.00022901860309680293, "epoch": 0.48, "step": 12850}
+{"loss": 0.7715, "learning_rate": 0.00022896290520218335, "epoch": 0.48, "step": 12860}
+{"loss": 0.7475, "learning_rate": 0.00022890720730756376, "epoch": 0.48, "step": 12870}
+{"loss": 0.6232, "learning_rate": 0.00022885150941294417, "epoch": 0.48, "step": 12880}
+{"loss": 0.6331, "learning_rate": 0.0002287958115183246, "epoch": 0.48, "step": 12890}
+{"loss": 0.6767, "learning_rate": 0.00022874011362370502, "epoch": 0.48, "step": 12900}
+{"loss": 0.7686, "learning_rate": 0.00022868441572908543, "epoch": 0.48, "step": 12910}
+{"loss": 0.6391, "learning_rate": 0.00022862871783446584, "epoch": 0.48, "step": 12920}
+{"loss": 0.7066, "learning_rate": 0.00022857301993984625, "epoch": 0.48, "step": 12930}
+{"loss": 0.6514, "learning_rate": 0.00022851732204522666, "epoch": 0.48, "step": 12940}
+{"loss": 0.7157, "learning_rate": 0.0002284616241506071, "epoch": 0.48, "step": 12950}
+{"loss": 0.6704, "learning_rate": 0.0002284059262559875, "epoch": 0.48, "step": 12960}
+{"loss": 0.6819, "learning_rate": 0.00022835579815082987, "epoch": 0.48, "step": 12970}
+{"loss": 0.7704, "learning_rate": 0.00022830010025621028, "epoch": 0.48, "step": 12980}
+{"loss": 0.7403, "learning_rate": 0.00022824440236159072, "epoch": 0.48, "step": 12990}
+{"loss": 0.716, "learning_rate": 0.00022818870446697113, "epoch": 0.48, "step": 13000}
+{"loss": 0.5772, "learning_rate": 0.00022813300657235154, "epoch": 0.48, "step": 13010}
+{"loss": 0.6466, "learning_rate": 0.00022807730867773195, "epoch": 0.48, "step": 13020}
+{"loss": 0.5935, "learning_rate": 0.00022802161078311236, "epoch": 0.48, "step": 13030}
+{"loss": 0.7362, "learning_rate": 0.00022796591288849278, "epoch": 0.48, "step": 13040}
+{"loss": 0.7085, "learning_rate": 0.0002279102149938732, "epoch": 0.48, "step": 13050}
+{"loss": 0.7303, "learning_rate": 0.00022785451709925365, "epoch": 0.48, "step": 13060}
+{"loss": 0.6836, "learning_rate": 0.00022779881920463406, "epoch": 0.48, "step": 13070}
+{"loss": 0.6263, "learning_rate": 0.00022774312131001447, "epoch": 0.48, "step": 13080}
+{"loss": 0.6741, "learning_rate": 0.00022768742341539489, "epoch": 0.49, "step": 13090}
+{"loss": 0.6999, "learning_rate": 0.0002276317255207753, "epoch": 0.49, "step": 13100}
+{"loss": 0.6771, "learning_rate": 0.0002275760276261557, "epoch": 0.49, "step": 13110}
+{"loss": 0.7091, "learning_rate": 0.00022752032973153615, "epoch": 0.49, "step": 13120}
+{"loss": 0.6451, "learning_rate": 0.00022746463183691656, "epoch": 0.49, "step": 13130}
+{"loss": 0.7333, "learning_rate": 0.00022740893394229697, "epoch": 0.49, "step": 13140}
+{"loss": 0.69, "learning_rate": 0.00022735323604767738, "epoch": 0.49, "step": 13150}
+{"loss": 0.775, "learning_rate": 0.0002272975381530578, "epoch": 0.49, "step": 13160}
+{"loss": 0.7927, "learning_rate": 0.0002272418402584382, "epoch": 0.49, "step": 13170}
+{"loss": 0.7457, "learning_rate": 0.0002271861423638186, "epoch": 0.49, "step": 13180}
+{"loss": 0.7266, "learning_rate": 0.00022713044446919905, "epoch": 0.49, "step": 13190}
+{"loss": 0.5273, "learning_rate": 0.00022707474657457946, "epoch": 0.49, "step": 13200}
+{"loss": 0.6495, "learning_rate": 0.00022701904867995987, "epoch": 0.49, "step": 13210}
+{"loss": 0.7933, "learning_rate": 0.00022696335078534028, "epoch": 0.49, "step": 13220}
+{"loss": 0.6314, "learning_rate": 0.0002269076528907207, "epoch": 0.49, "step": 13230}
+{"loss": 0.7547, "learning_rate": 0.0002268519549961011, "epoch": 0.49, "step": 13240}
+{"loss": 0.716, "learning_rate": 0.00022679625710148157, "epoch": 0.49, "step": 13250}
+{"loss": 0.8462, "learning_rate": 0.00022674055920686198, "epoch": 0.49, "step": 13260}
+{"loss": 0.8108, "learning_rate": 0.0002266848613122424, "epoch": 0.49, "step": 13270}
+{"loss": 0.5796, "learning_rate": 0.0002266291634176228, "epoch": 0.49, "step": 13280}
+{"loss": 0.6584, "learning_rate": 0.00022657346552300322, "epoch": 0.49, "step": 13290}
+{"loss": 0.7407, "learning_rate": 0.00022651776762838363, "epoch": 0.49, "step": 13300}
+{"loss": 0.6505, "learning_rate": 0.00022646206973376404, "epoch": 0.49, "step": 13310}
+{"loss": 0.6075, "learning_rate": 0.00022640637183914448, "epoch": 0.49, "step": 13320}
+{"loss": 0.6258, "learning_rate": 0.0002263506739445249, "epoch": 0.49, "step": 13330}
+{"loss": 0.5988, "learning_rate": 0.0002262949760499053, "epoch": 0.49, "step": 13340}
+{"loss": 0.7777, "learning_rate": 0.0002262392781552857, "epoch": 0.49, "step": 13350}
+{"loss": 0.763, "learning_rate": 0.00022618358026066612, "epoch": 0.5, "step": 13360}
+{"loss": 0.7065, "learning_rate": 0.00022612788236604653, "epoch": 0.5, "step": 13370}
+{"loss": 0.7195, "learning_rate": 0.00022607218447142697, "epoch": 0.5, "step": 13380}
+{"loss": 0.5759, "learning_rate": 0.00022601648657680738, "epoch": 0.5, "step": 13390}
+{"loss": 0.727, "learning_rate": 0.0002259607886821878, "epoch": 0.5, "step": 13400}
+{"loss": 0.5891, "learning_rate": 0.0002259050907875682, "epoch": 0.5, "step": 13410}
+{"loss": 0.565, "learning_rate": 0.00022584939289294862, "epoch": 0.5, "step": 13420}
+{"loss": 0.655, "learning_rate": 0.00022579369499832903, "epoch": 0.5, "step": 13430}
+{"loss": 0.7448, "learning_rate": 0.00022573799710370944, "epoch": 0.5, "step": 13440}
+{"loss": 0.7091, "learning_rate": 0.00022568229920908988, "epoch": 0.5, "step": 13450}
+{"loss": 0.6339, "learning_rate": 0.00022562660131447031, "epoch": 0.5, "step": 13460}
+{"loss": 0.7307, "learning_rate": 0.00022557090341985073, "epoch": 0.5, "step": 13470}
+{"loss": 0.6675, "learning_rate": 0.00022551520552523114, "epoch": 0.5, "step": 13480}
+{"loss": 0.7426, "learning_rate": 0.00022545950763061155, "epoch": 0.5, "step": 13490}
+{"loss": 0.7154, "learning_rate": 0.00022540380973599196, "epoch": 0.5, "step": 13500}
+{"loss": 0.6243, "learning_rate": 0.0002253481118413724, "epoch": 0.5, "step": 13510}
+{"loss": 0.5902, "learning_rate": 0.0002252924139467528, "epoch": 0.5, "step": 13520}
+{"loss": 0.6257, "learning_rate": 0.00022523671605213322, "epoch": 0.5, "step": 13530}
+{"loss": 0.6915, "learning_rate": 0.00022518101815751363, "epoch": 0.5, "step": 13540}
+{"loss": 0.6206, "learning_rate": 0.00022512532026289404, "epoch": 0.5, "step": 13550}
+{"loss": 0.5492, "learning_rate": 0.00022506962236827445, "epoch": 0.5, "step": 13560}
+{"loss": 0.7209, "learning_rate": 0.00022501392447365486, "epoch": 0.5, "step": 13570}
+{"loss": 0.6182, "learning_rate": 0.0002249582265790353, "epoch": 0.5, "step": 13580}
+{"loss": 0.7192, "learning_rate": 0.0002249025286844157, "epoch": 0.5, "step": 13590}
+{"loss": 0.7341, "learning_rate": 0.00022484683078979612, "epoch": 0.5, "step": 13600}
+{"loss": 0.732, "learning_rate": 0.00022479113289517654, "epoch": 0.5, "step": 13610}
+{"loss": 0.6638, "learning_rate": 0.00022473543500055695, "epoch": 0.5, "step": 13620}
+{"loss": 0.4943, "learning_rate": 0.00022467973710593736, "epoch": 0.51, "step": 13630}
+{"loss": 0.7328, "learning_rate": 0.0002246240392113178, "epoch": 0.51, "step": 13640}
+{"loss": 0.7483, "learning_rate": 0.0002245683413166982, "epoch": 0.51, "step": 13650}
+{"loss": 0.6602, "learning_rate": 0.00022451264342207862, "epoch": 0.51, "step": 13660}
+{"loss": 0.7906, "learning_rate": 0.00022445694552745903, "epoch": 0.51, "step": 13670}
+{"loss": 0.6396, "learning_rate": 0.00022440124763283944, "epoch": 0.51, "step": 13680}
+{"loss": 0.639, "learning_rate": 0.00022434554973821985, "epoch": 0.51, "step": 13690}
+{"loss": 0.6533, "learning_rate": 0.0002242898518436003, "epoch": 0.51, "step": 13700}
+{"loss": 0.694, "learning_rate": 0.00022423415394898073, "epoch": 0.51, "step": 13710}
+{"loss": 0.8023, "learning_rate": 0.00022417845605436114, "epoch": 0.51, "step": 13720}
+{"loss": 0.6403, "learning_rate": 0.00022412275815974155, "epoch": 0.51, "step": 13730}
+{"loss": 0.7514, "learning_rate": 0.00022406706026512196, "epoch": 0.51, "step": 13740}
+{"loss": 0.7525, "learning_rate": 0.00022401136237050237, "epoch": 0.51, "step": 13750}
+{"loss": 0.7266, "learning_rate": 0.00022395566447588278, "epoch": 0.51, "step": 13760}
+{"loss": 0.6511, "learning_rate": 0.00022389996658126322, "epoch": 0.51, "step": 13770}
+{"loss": 0.5995, "learning_rate": 0.00022384426868664363, "epoch": 0.51, "step": 13780}
+{"loss": 0.6895, "learning_rate": 0.00022378857079202404, "epoch": 0.51, "step": 13790}
+{"loss": 0.623, "learning_rate": 0.00022373287289740446, "epoch": 0.51, "step": 13800}
+{"loss": 0.6552, "learning_rate": 0.00022367717500278487, "epoch": 0.51, "step": 13810}
+{"loss": 0.7724, "learning_rate": 0.00022362147710816528, "epoch": 0.51, "step": 13820}
+{"loss": 0.6573, "learning_rate": 0.0002235657792135457, "epoch": 0.51, "step": 13830}
+{"loss": 0.656, "learning_rate": 0.00022351008131892613, "epoch": 0.51, "step": 13840}
+{"loss": 0.6983, "learning_rate": 0.00022345438342430654, "epoch": 0.51, "step": 13850}
+{"loss": 0.7186, "learning_rate": 0.00022339868552968695, "epoch": 0.51, "step": 13860}
+{"loss": 0.7065, "learning_rate": 0.00022334298763506736, "epoch": 0.51, "step": 13870}
+{"loss": 0.6109, "learning_rate": 0.00022328728974044777, "epoch": 0.51, "step": 13880}
+{"loss": 0.717, "learning_rate": 0.00022323159184582818, "epoch": 0.51, "step": 13890}
+{"loss": 0.6701, "learning_rate": 0.00022317589395120865, "epoch": 0.52, "step": 13900}
+{"loss": 0.6823, "learning_rate": 0.00022312019605658906, "epoch": 0.52, "step": 13910}
+{"loss": 0.4733, "learning_rate": 0.00022306449816196947, "epoch": 0.52, "step": 13920}
+{"loss": 0.773, "learning_rate": 0.00022300880026734988, "epoch": 0.52, "step": 13930}
+{"loss": 0.6302, "learning_rate": 0.0002229531023727303, "epoch": 0.52, "step": 13940}
+{"loss": 0.6834, "learning_rate": 0.0002228974044781107, "epoch": 0.52, "step": 13950}
+{"loss": 0.7129, "learning_rate": 0.00022284170658349112, "epoch": 0.52, "step": 13960}
+{"loss": 0.6505, "learning_rate": 0.00022278600868887155, "epoch": 0.52, "step": 13970}
+{"loss": 0.6433, "learning_rate": 0.00022273031079425196, "epoch": 0.52, "step": 13980}
+{"loss": 0.7738, "learning_rate": 0.00022267461289963238, "epoch": 0.52, "step": 13990}
+{"loss": 0.6346, "learning_rate": 0.0002226189150050128, "epoch": 0.52, "step": 14000}
+{"loss": 0.6472, "learning_rate": 0.0002225632171103932, "epoch": 0.52, "step": 14010}
+{"loss": 0.8198, "learning_rate": 0.0002225075192157736, "epoch": 0.52, "step": 14020}
+{"loss": 0.7712, "learning_rate": 0.00022245182132115405, "epoch": 0.52, "step": 14030}
+{"loss": 0.614, "learning_rate": 0.00022239612342653446, "epoch": 0.52, "step": 14040}
+{"loss": 0.6847, "learning_rate": 0.00022234042553191487, "epoch": 0.52, "step": 14050}
+{"loss": 0.5931, "learning_rate": 0.00022228472763729528, "epoch": 0.52, "step": 14060}
+{"loss": 0.5816, "learning_rate": 0.0002222290297426757, "epoch": 0.52, "step": 14070}
+{"loss": 0.6594, "learning_rate": 0.0002221733318480561, "epoch": 0.52, "step": 14080}
+{"loss": 0.6311, "learning_rate": 0.00022211763395343651, "epoch": 0.52, "step": 14090}
+{"loss": 0.7299, "learning_rate": 0.00022206193605881698, "epoch": 0.52, "step": 14100}
+{"loss": 0.6198, "learning_rate": 0.0002220062381641974, "epoch": 0.52, "step": 14110}
+{"loss": 0.8008, "learning_rate": 0.0002219505402695778, "epoch": 0.52, "step": 14120}
+{"loss": 0.6623, "learning_rate": 0.0002218948423749582, "epoch": 0.52, "step": 14130}
+{"loss": 0.6649, "learning_rate": 0.00022183914448033862, "epoch": 0.52, "step": 14140}
+{"loss": 0.6897, "learning_rate": 0.00022178344658571904, "epoch": 0.52, "step": 14150}
+{"loss": 0.5862, "learning_rate": 0.00022172774869109947, "epoch": 0.52, "step": 14160}
+{"loss": 0.6233, "learning_rate": 0.00022167205079647988, "epoch": 0.53, "step": 14170}
+{"loss": 0.6554, "learning_rate": 0.0002216163529018603, "epoch": 0.53, "step": 14180}
+{"loss": 0.6932, "learning_rate": 0.0002215606550072407, "epoch": 0.53, "step": 14190}
+{"loss": 0.719, "learning_rate": 0.00022150495711262112, "epoch": 0.53, "step": 14200}
+{"loss": 0.6083, "learning_rate": 0.00022144925921800153, "epoch": 0.53, "step": 14210}
+{"loss": 0.6489, "learning_rate": 0.00022139356132338194, "epoch": 0.53, "step": 14220}
+{"loss": 0.6216, "learning_rate": 0.00022133786342876238, "epoch": 0.53, "step": 14230}
+{"loss": 0.7405, "learning_rate": 0.0002212821655341428, "epoch": 0.53, "step": 14240}
+{"loss": 0.6268, "learning_rate": 0.0002212264676395232, "epoch": 0.53, "step": 14250}
+{"loss": 0.7664, "learning_rate": 0.0002211707697449036, "epoch": 0.53, "step": 14260}
+{"loss": 0.8237, "learning_rate": 0.00022111507185028402, "epoch": 0.53, "step": 14270}
+{"loss": 0.7146, "learning_rate": 0.00022105937395566443, "epoch": 0.53, "step": 14280}
+{"loss": 0.6834, "learning_rate": 0.0002210036760610449, "epoch": 0.53, "step": 14290}
+{"loss": 0.7013, "learning_rate": 0.0002209479781664253, "epoch": 0.53, "step": 14300}
+{"loss": 0.6318, "learning_rate": 0.00022089228027180572, "epoch": 0.53, "step": 14310}
+{"loss": 0.6377, "learning_rate": 0.00022083658237718613, "epoch": 0.53, "step": 14320}
+{"loss": 0.6101, "learning_rate": 0.00022078088448256654, "epoch": 0.53, "step": 14330}
+{"loss": 0.5967, "learning_rate": 0.00022072518658794696, "epoch": 0.53, "step": 14340}
+{"loss": 0.5797, "learning_rate": 0.00022066948869332737, "epoch": 0.53, "step": 14350}
+{"loss": 0.7185, "learning_rate": 0.0002206137907987078, "epoch": 0.53, "step": 14360}
+{"loss": 0.6422, "learning_rate": 0.00022055809290408822, "epoch": 0.53, "step": 14370}
+{"loss": 0.5973, "learning_rate": 0.00022050239500946863, "epoch": 0.53, "step": 14380}
+{"loss": 0.7087, "learning_rate": 0.00022044669711484904, "epoch": 0.53, "step": 14390}
+{"loss": 0.7315, "learning_rate": 0.00022039099922022945, "epoch": 0.53, "step": 14400}
+{"loss": 0.5749, "learning_rate": 0.00022033530132560986, "epoch": 0.53, "step": 14410}
+{"loss": 0.6827, "learning_rate": 0.0002202796034309903, "epoch": 0.53, "step": 14420}
+{"loss": 0.6614, "learning_rate": 0.0002202239055363707, "epoch": 0.53, "step": 14430}
+{"loss": 0.7314, "learning_rate": 0.00022016820764175112, "epoch": 0.54, "step": 14440}
+{"loss": 0.7085, "learning_rate": 0.00022011250974713153, "epoch": 0.54, "step": 14450}
+{"loss": 0.7075, "learning_rate": 0.00022005681185251194, "epoch": 0.54, "step": 14460}
+{"loss": 0.659, "learning_rate": 0.00022000111395789235, "epoch": 0.54, "step": 14470}
+{"loss": 0.753, "learning_rate": 0.00021994541606327277, "epoch": 0.54, "step": 14480}
+{"loss": 0.7349, "learning_rate": 0.00021988971816865323, "epoch": 0.54, "step": 14490}
+{"loss": 0.7157, "learning_rate": 0.00021983402027403364, "epoch": 0.54, "step": 14500}
+{"loss": 0.6433, "learning_rate": 0.00021977832237941405, "epoch": 0.54, "step": 14510}
+{"loss": 0.825, "learning_rate": 0.00021972262448479446, "epoch": 0.54, "step": 14520}
+{"loss": 0.7108, "learning_rate": 0.00021966692659017488, "epoch": 0.54, "step": 14530}
+{"loss": 0.7532, "learning_rate": 0.00021961122869555529, "epoch": 0.54, "step": 14540}
+{"loss": 0.5937, "learning_rate": 0.00021955553080093572, "epoch": 0.54, "step": 14550}
+{"loss": 0.7208, "learning_rate": 0.00021949983290631614, "epoch": 0.54, "step": 14560}
+{"loss": 0.6708, "learning_rate": 0.00021944413501169655, "epoch": 0.54, "step": 14570}
+{"loss": 0.7305, "learning_rate": 0.00021938843711707696, "epoch": 0.54, "step": 14580}
+{"loss": 0.6432, "learning_rate": 0.00021933273922245737, "epoch": 0.54, "step": 14590}
+{"loss": 0.7485, "learning_rate": 0.00021927704132783778, "epoch": 0.54, "step": 14600}
+{"loss": 0.6242, "learning_rate": 0.00021922134343321822, "epoch": 0.54, "step": 14610}
+{"loss": 0.6528, "learning_rate": 0.00021916564553859863, "epoch": 0.54, "step": 14620}
+{"loss": 0.693, "learning_rate": 0.00021910994764397904, "epoch": 0.54, "step": 14630}
+{"loss": 0.645, "learning_rate": 0.00021905424974935945, "epoch": 0.54, "step": 14640}
+{"loss": 0.8306, "learning_rate": 0.00021899855185473986, "epoch": 0.54, "step": 14650}
+{"loss": 0.657, "learning_rate": 0.00021894285396012027, "epoch": 0.54, "step": 14660}
+{"loss": 0.6826, "learning_rate": 0.00021888715606550069, "epoch": 0.54, "step": 14670}
+{"loss": 0.6073, "learning_rate": 0.00021883145817088112, "epoch": 0.54, "step": 14680}
+{"loss": 0.7809, "learning_rate": 0.00021877576027626153, "epoch": 0.54, "step": 14690}
+{"loss": 0.6255, "learning_rate": 0.00021872006238164195, "epoch": 0.54, "step": 14700}
+{"loss": 0.4938, "learning_rate": 0.00021866436448702236, "epoch": 0.55, "step": 14710}
+{"loss": 0.8137, "learning_rate": 0.00021860866659240277, "epoch": 0.55, "step": 14720}
+{"loss": 0.7493, "learning_rate": 0.0002185529686977832, "epoch": 0.55, "step": 14730}
+{"loss": 0.789, "learning_rate": 0.00021849727080316364, "epoch": 0.55, "step": 14740}
+{"loss": 0.7218, "learning_rate": 0.00021844157290854406, "epoch": 0.55, "step": 14750}
+{"loss": 0.7999, "learning_rate": 0.00021838587501392447, "epoch": 0.55, "step": 14760}
+{"loss": 0.6433, "learning_rate": 0.00021833017711930488, "epoch": 0.55, "step": 14770}
+{"loss": 0.692, "learning_rate": 0.0002182744792246853, "epoch": 0.55, "step": 14780}
+{"loss": 0.6873, "learning_rate": 0.0002182187813300657, "epoch": 0.55, "step": 14790}
+{"loss": 0.7035, "learning_rate": 0.0002181630834354461, "epoch": 0.55, "step": 14800}
+{"loss": 0.641, "learning_rate": 0.00021810738554082655, "epoch": 0.55, "step": 14810}
+{"loss": 0.6117, "learning_rate": 0.00021805168764620696, "epoch": 0.55, "step": 14820}
+{"loss": 0.6816, "learning_rate": 0.00021799598975158737, "epoch": 0.55, "step": 14830}
+{"loss": 0.643, "learning_rate": 0.00021794029185696778, "epoch": 0.55, "step": 14840}
+{"loss": 0.6706, "learning_rate": 0.0002178845939623482, "epoch": 0.55, "step": 14850}
+{"loss": 0.629, "learning_rate": 0.0002178288960677286, "epoch": 0.55, "step": 14860}
+{"loss": 0.6919, "learning_rate": 0.00021777319817310904, "epoch": 0.55, "step": 14870}
+{"loss": 0.6551, "learning_rate": 0.00021771750027848945, "epoch": 0.55, "step": 14880}
+{"loss": 0.8308, "learning_rate": 0.00021766180238386987, "epoch": 0.55, "step": 14890}
+{"loss": 0.5696, "learning_rate": 0.00021760610448925028, "epoch": 0.55, "step": 14900}
+{"loss": 0.663, "learning_rate": 0.0002175504065946307, "epoch": 0.55, "step": 14910}
+{"loss": 0.5983, "learning_rate": 0.0002174947087000111, "epoch": 0.55, "step": 14920}
+{"loss": 0.6473, "learning_rate": 0.0002174390108053915, "epoch": 0.55, "step": 14930}
+{"loss": 0.6577, "learning_rate": 0.00021738331291077198, "epoch": 0.55, "step": 14940}
+{"loss": 0.7108, "learning_rate": 0.0002173276150161524, "epoch": 0.55, "step": 14950}
+{"loss": 0.6457, "learning_rate": 0.0002172719171215328, "epoch": 0.55, "step": 14960}
+{"loss": 0.5915, "learning_rate": 0.0002172162192269132, "epoch": 0.55, "step": 14970}
+{"loss": 0.8125, "learning_rate": 0.00021716052133229362, "epoch": 0.56, "step": 14980}
+{"loss": 0.6549, "learning_rate": 0.00021710482343767403, "epoch": 0.56, "step": 14990}
+{"loss": 0.6892, "learning_rate": 0.00021704912554305447, "epoch": 0.56, "step": 15000}
+{"loss": 0.5753, "learning_rate": 0.00021699342764843488, "epoch": 0.56, "step": 15010}
+{"loss": 0.6894, "learning_rate": 0.0002169377297538153, "epoch": 0.56, "step": 15020}
+{"loss": 0.7092, "learning_rate": 0.0002168820318591957, "epoch": 0.56, "step": 15030}
+{"loss": 0.6203, "learning_rate": 0.00021682633396457611, "epoch": 0.56, "step": 15040}
+{"loss": 0.7342, "learning_rate": 0.00021677063606995653, "epoch": 0.56, "step": 15050}
+{"loss": 0.7339, "learning_rate": 0.00021671493817533694, "epoch": 0.56, "step": 15060}
+{"loss": 0.6012, "learning_rate": 0.00021665924028071737, "epoch": 0.56, "step": 15070}
+{"loss": 0.6904, "learning_rate": 0.00021660354238609779, "epoch": 0.56, "step": 15080}
+{"loss": 0.6205, "learning_rate": 0.0002165478444914782, "epoch": 0.56, "step": 15090}
+{"loss": 0.5442, "learning_rate": 0.0002164921465968586, "epoch": 0.56, "step": 15100}
+{"loss": 0.7029, "learning_rate": 0.00021643644870223902, "epoch": 0.56, "step": 15110}
+{"loss": 0.7041, "learning_rate": 0.00021638075080761943, "epoch": 0.56, "step": 15120}
+{"loss": 0.7117, "learning_rate": 0.0002163250529129999, "epoch": 0.56, "step": 15130}
+{"loss": 0.6666, "learning_rate": 0.0002162693550183803, "epoch": 0.56, "step": 15140}
+{"loss": 0.672, "learning_rate": 0.00021621365712376072, "epoch": 0.56, "step": 15150}
+{"loss": 0.6775, "learning_rate": 0.00021615795922914113, "epoch": 0.56, "step": 15160}
+{"loss": 0.6723, "learning_rate": 0.00021610226133452154, "epoch": 0.56, "step": 15170}
+{"loss": 0.6951, "learning_rate": 0.00021604656343990195, "epoch": 0.56, "step": 15180}
+{"loss": 0.6959, "learning_rate": 0.00021599086554528236, "epoch": 0.56, "step": 15190}
+{"loss": 0.7746, "learning_rate": 0.0002159351676506628, "epoch": 0.56, "step": 15200}
+{"loss": 0.6634, "learning_rate": 0.0002158794697560432, "epoch": 0.56, "step": 15210}
+{"loss": 0.6506, "learning_rate": 0.00021582377186142362, "epoch": 0.56, "step": 15220}
+{"loss": 0.7388, "learning_rate": 0.00021576807396680403, "epoch": 0.56, "step": 15230}
+{"loss": 0.581, "learning_rate": 0.00021571237607218445, "epoch": 0.56, "step": 15240}
+{"loss": 0.6209, "learning_rate": 0.00021565667817756486, "epoch": 0.57, "step": 15250}
+{"loss": 0.6746, "learning_rate": 0.0002156009802829453, "epoch": 0.57, "step": 15260}
+{"loss": 0.6239, "learning_rate": 0.0002155452823883257, "epoch": 0.57, "step": 15270}
+{"loss": 0.7832, "learning_rate": 0.00021548958449370612, "epoch": 0.57, "step": 15280}
+{"loss": 0.6331, "learning_rate": 0.00021543388659908653, "epoch": 0.57, "step": 15290}
+{"loss": 0.6882, "learning_rate": 0.00021537818870446694, "epoch": 0.57, "step": 15300}
+{"loss": 0.7722, "learning_rate": 0.00021532249080984735, "epoch": 0.57, "step": 15310}
+{"loss": 0.6875, "learning_rate": 0.00021526679291522776, "epoch": 0.57, "step": 15320}
+{"loss": 0.5636, "learning_rate": 0.00021521109502060823, "epoch": 0.57, "step": 15330}
+{"loss": 0.7001, "learning_rate": 0.00021515539712598864, "epoch": 0.57, "step": 15340}
+{"loss": 0.5869, "learning_rate": 0.00021509969923136905, "epoch": 0.57, "step": 15350}
+{"loss": 0.6365, "learning_rate": 0.00021504400133674946, "epoch": 0.57, "step": 15360}
+{"loss": 0.7286, "learning_rate": 0.00021499387323159185, "epoch": 0.57, "step": 15370}
+{"loss": 0.6721, "learning_rate": 0.00021493817533697226, "epoch": 0.57, "step": 15380}
+{"loss": 0.5865, "learning_rate": 0.00021488247744235267, "epoch": 0.57, "step": 15390}
+{"loss": 0.6176, "learning_rate": 0.00021482677954773308, "epoch": 0.57, "step": 15400}
+{"loss": 0.756, "learning_rate": 0.0002147710816531135, "epoch": 0.57, "step": 15410}
+{"loss": 0.7346, "learning_rate": 0.0002147153837584939, "epoch": 0.57, "step": 15420}
+{"loss": 0.5988, "learning_rate": 0.00021465968586387434, "epoch": 0.57, "step": 15430}
+{"loss": 0.7183, "learning_rate": 0.00021460398796925475, "epoch": 0.57, "step": 15440}
+{"loss": 0.7797, "learning_rate": 0.00021454829007463516, "epoch": 0.57, "step": 15450}
+{"loss": 0.593, "learning_rate": 0.00021449259218001557, "epoch": 0.57, "step": 15460}
+{"loss": 0.6311, "learning_rate": 0.00021443689428539598, "epoch": 0.57, "step": 15470}
+{"loss": 0.5723, "learning_rate": 0.0002143811963907764, "epoch": 0.57, "step": 15480}
+{"loss": 0.7624, "learning_rate": 0.0002143254984961568, "epoch": 0.57, "step": 15490}
+{"loss": 0.6886, "learning_rate": 0.00021426980060153725, "epoch": 0.57, "step": 15500}
+{"loss": 0.7613, "learning_rate": 0.00021421410270691766, "epoch": 0.57, "step": 15510}
+{"loss": 0.7022, "learning_rate": 0.00021415840481229807, "epoch": 0.58, "step": 15520}
+{"loss": 0.6362, "learning_rate": 0.00021410270691767848, "epoch": 0.58, "step": 15530}
+{"loss": 0.6311, "learning_rate": 0.0002140470090230589, "epoch": 0.58, "step": 15540}
+{"loss": 0.6845, "learning_rate": 0.0002139913111284393, "epoch": 0.58, "step": 15550}
+{"loss": 0.5831, "learning_rate": 0.00021393561323381977, "epoch": 0.58, "step": 15560}
+{"loss": 0.6478, "learning_rate": 0.00021387991533920018, "epoch": 0.58, "step": 15570}
+{"loss": 0.7938, "learning_rate": 0.0002138242174445806, "epoch": 0.58, "step": 15580}
+{"loss": 0.6994, "learning_rate": 0.000213768519549961, "epoch": 0.58, "step": 15590}
+{"loss": 0.6496, "learning_rate": 0.0002137128216553414, "epoch": 0.58, "step": 15600}
+{"loss": 0.7755, "learning_rate": 0.00021365712376072182, "epoch": 0.58, "step": 15610}
+{"loss": 0.6232, "learning_rate": 0.00021360142586610223, "epoch": 0.58, "step": 15620}
+{"loss": 0.7156, "learning_rate": 0.00021354572797148267, "epoch": 0.58, "step": 15630}
+{"loss": 0.8258, "learning_rate": 0.00021349003007686308, "epoch": 0.58, "step": 15640}
+{"loss": 0.6521, "learning_rate": 0.0002134343321822435, "epoch": 0.58, "step": 15650}
+{"loss": 0.6671, "learning_rate": 0.0002133786342876239, "epoch": 0.58, "step": 15660}
+{"loss": 0.6472, "learning_rate": 0.00021332293639300432, "epoch": 0.58, "step": 15670}
+{"loss": 0.7124, "learning_rate": 0.00021326723849838473, "epoch": 0.58, "step": 15680}
+{"loss": 0.4767, "learning_rate": 0.00021321154060376517, "epoch": 0.58, "step": 15690}
+{"loss": 0.5941, "learning_rate": 0.00021315584270914558, "epoch": 0.58, "step": 15700}
+{"loss": 0.584, "learning_rate": 0.000213100144814526, "epoch": 0.58, "step": 15710}
+{"loss": 0.7129, "learning_rate": 0.0002130444469199064, "epoch": 0.58, "step": 15720}
+{"loss": 0.6647, "learning_rate": 0.0002129887490252868, "epoch": 0.58, "step": 15730}
+{"loss": 0.6782, "learning_rate": 0.00021293305113066722, "epoch": 0.58, "step": 15740}
+{"loss": 0.7348, "learning_rate": 0.00021287735323604763, "epoch": 0.58, "step": 15750}
+{"loss": 0.5956, "learning_rate": 0.0002128216553414281, "epoch": 0.58, "step": 15760}
+{"loss": 0.7793, "learning_rate": 0.0002127659574468085, "epoch": 0.58, "step": 15770}
+{"loss": 0.7324, "learning_rate": 0.00021271025955218892, "epoch": 0.58, "step": 15780}
+{"loss": 0.6776, "learning_rate": 0.00021265456165756933, "epoch": 0.59, "step": 15790}
+{"loss": 0.6559, "learning_rate": 0.00021259886376294974, "epoch": 0.59, "step": 15800}
+{"loss": 0.6334, "learning_rate": 0.00021254316586833015, "epoch": 0.59, "step": 15810}
+{"loss": 0.6819, "learning_rate": 0.0002124874679737106, "epoch": 0.59, "step": 15820}
+{"loss": 0.6798, "learning_rate": 0.000212431770079091, "epoch": 0.59, "step": 15830}
+{"loss": 0.5819, "learning_rate": 0.00021237607218447141, "epoch": 0.59, "step": 15840}
+{"loss": 0.6284, "learning_rate": 0.00021232037428985182, "epoch": 0.59, "step": 15850}
+{"loss": 0.8205, "learning_rate": 0.00021226467639523224, "epoch": 0.59, "step": 15860}
+{"loss": 0.6568, "learning_rate": 0.00021220897850061265, "epoch": 0.59, "step": 15870}
+{"loss": 0.6719, "learning_rate": 0.00021215328060599306, "epoch": 0.59, "step": 15880}
+{"loss": 0.6046, "learning_rate": 0.0002120975827113735, "epoch": 0.59, "step": 15890}
+{"loss": 0.6394, "learning_rate": 0.0002120418848167539, "epoch": 0.59, "step": 15900}
+{"loss": 0.6688, "learning_rate": 0.00021198618692213432, "epoch": 0.59, "step": 15910}
+{"loss": 0.5539, "learning_rate": 0.00021193048902751473, "epoch": 0.59, "step": 15920}
+{"loss": 0.5964, "learning_rate": 0.00021187479113289514, "epoch": 0.59, "step": 15930}
+{"loss": 0.6626, "learning_rate": 0.00021181909323827555, "epoch": 0.59, "step": 15940}
+{"loss": 0.6718, "learning_rate": 0.00021176339534365602, "epoch": 0.59, "step": 15950}
+{"loss": 0.6036, "learning_rate": 0.00021170769744903643, "epoch": 0.59, "step": 15960}
+{"loss": 0.673, "learning_rate": 0.00021165199955441684, "epoch": 0.59, "step": 15970}
+{"loss": 0.5856, "learning_rate": 0.00021159630165979725, "epoch": 0.59, "step": 15980}
+{"loss": 0.6437, "learning_rate": 0.00021154060376517766, "epoch": 0.59, "step": 15990}
+{"loss": 0.5944, "learning_rate": 0.00021148490587055807, "epoch": 0.59, "step": 16000}
+{"loss": 0.7816, "learning_rate": 0.00021142920797593848, "epoch": 0.59, "step": 16010}
+{"loss": 0.5892, "learning_rate": 0.00021137351008131892, "epoch": 0.59, "step": 16020}
+{"loss": 0.591, "learning_rate": 0.00021131781218669933, "epoch": 0.59, "step": 16030}
+{"loss": 0.6353, "learning_rate": 0.00021126211429207974, "epoch": 0.59, "step": 16040}
+{"loss": 0.6361, "learning_rate": 0.00021120641639746016, "epoch": 0.59, "step": 16050}
+{"loss": 0.7992, "learning_rate": 0.00021115071850284057, "epoch": 0.6, "step": 16060}
+{"loss": 0.6903, "learning_rate": 0.00021109502060822098, "epoch": 0.6, "step": 16070}
+{"loss": 0.6197, "learning_rate": 0.00021103932271360142, "epoch": 0.6, "step": 16080}
+{"loss": 0.7122, "learning_rate": 0.00021098362481898183, "epoch": 0.6, "step": 16090}
+{"loss": 0.5584, "learning_rate": 0.00021092792692436224, "epoch": 0.6, "step": 16100}
+{"loss": 0.637, "learning_rate": 0.00021087222902974265, "epoch": 0.6, "step": 16110}
+{"loss": 0.7129, "learning_rate": 0.00021081653113512306, "epoch": 0.6, "step": 16120}
+{"loss": 0.5697, "learning_rate": 0.00021076083324050347, "epoch": 0.6, "step": 16130}
+{"loss": 0.7463, "learning_rate": 0.00021070513534588388, "epoch": 0.6, "step": 16140}
+{"loss": 0.6301, "learning_rate": 0.00021064943745126435, "epoch": 0.6, "step": 16150}
+{"loss": 0.8484, "learning_rate": 0.00021059373955664476, "epoch": 0.6, "step": 16160}
+{"loss": 0.5577, "learning_rate": 0.00021053804166202517, "epoch": 0.6, "step": 16170}
+{"loss": 0.7574, "learning_rate": 0.00021048234376740558, "epoch": 0.6, "step": 16180}
+{"loss": 0.6606, "learning_rate": 0.000210426645872786, "epoch": 0.6, "step": 16190}
+{"loss": 0.7742, "learning_rate": 0.0002103709479781664, "epoch": 0.6, "step": 16200}
+{"loss": 0.7174, "learning_rate": 0.00021031525008354684, "epoch": 0.6, "step": 16210}
+{"loss": 0.6275, "learning_rate": 0.00021025955218892725, "epoch": 0.6, "step": 16220}
+{"loss": 0.7237, "learning_rate": 0.00021020385429430766, "epoch": 0.6, "step": 16230}
+{"loss": 0.7215, "learning_rate": 0.00021014815639968808, "epoch": 0.6, "step": 16240}
+{"loss": 0.5755, "learning_rate": 0.0002100924585050685, "epoch": 0.6, "step": 16250}
+{"loss": 0.5667, "learning_rate": 0.0002100367606104489, "epoch": 0.6, "step": 16260}
+{"loss": 0.6133, "learning_rate": 0.0002099810627158293, "epoch": 0.6, "step": 16270}
+{"loss": 0.7582, "learning_rate": 0.00020992536482120975, "epoch": 0.6, "step": 16280}
+{"loss": 0.6751, "learning_rate": 0.00020986966692659016, "epoch": 0.6, "step": 16290}
+{"loss": 0.6342, "learning_rate": 0.00020981396903197057, "epoch": 0.6, "step": 16300}
+{"loss": 0.6421, "learning_rate": 0.00020975827113735098, "epoch": 0.6, "step": 16310}
+{"loss": 0.6766, "learning_rate": 0.0002097025732427314, "epoch": 0.6, "step": 16320}
+{"loss": 0.6065, "learning_rate": 0.0002096468753481118, "epoch": 0.61, "step": 16330}
+{"loss": 0.7122, "learning_rate": 0.00020959117745349224, "epoch": 0.61, "step": 16340}
+{"loss": 0.6652, "learning_rate": 0.00020953547955887265, "epoch": 0.61, "step": 16350}
+{"loss": 0.7515, "learning_rate": 0.00020947978166425306, "epoch": 0.61, "step": 16360}
+{"loss": 0.7357, "learning_rate": 0.00020942408376963348, "epoch": 0.61, "step": 16370}
+{"loss": 0.7068, "learning_rate": 0.0002093683858750139, "epoch": 0.61, "step": 16380}
+{"loss": 0.6725, "learning_rate": 0.00020931268798039432, "epoch": 0.61, "step": 16390}
+{"loss": 0.6435, "learning_rate": 0.00020925699008577474, "epoch": 0.61, "step": 16400}
+{"loss": 0.8076, "learning_rate": 0.00020920129219115517, "epoch": 0.61, "step": 16410}
+{"loss": 0.6132, "learning_rate": 0.00020914559429653558, "epoch": 0.61, "step": 16420}
+{"loss": 0.5879, "learning_rate": 0.000209089896401916, "epoch": 0.61, "step": 16430}
+{"loss": 0.6298, "learning_rate": 0.0002090341985072964, "epoch": 0.61, "step": 16440}
+{"loss": 0.6888, "learning_rate": 0.00020897850061267682, "epoch": 0.61, "step": 16450}
+{"loss": 0.7021, "learning_rate": 0.00020892280271805723, "epoch": 0.61, "step": 16460}
+{"loss": 0.7169, "learning_rate": 0.00020886710482343767, "epoch": 0.61, "step": 16470}
+{"loss": 0.6699, "learning_rate": 0.00020881140692881808, "epoch": 0.61, "step": 16480}
+{"loss": 0.5833, "learning_rate": 0.0002087557090341985, "epoch": 0.61, "step": 16490}
+{"loss": 0.7712, "learning_rate": 0.0002087000111395789, "epoch": 0.61, "step": 16500}
+{"loss": 0.6582, "learning_rate": 0.0002086443132449593, "epoch": 0.61, "step": 16510}
+{"loss": 0.7592, "learning_rate": 0.00020858861535033972, "epoch": 0.61, "step": 16520}
+{"loss": 0.7448, "learning_rate": 0.00020853291745572016, "epoch": 0.61, "step": 16530}
+{"loss": 0.7102, "learning_rate": 0.00020847721956110057, "epoch": 0.61, "step": 16540}
+{"loss": 0.5721, "learning_rate": 0.00020842152166648098, "epoch": 0.61, "step": 16550}
+{"loss": 0.6425, "learning_rate": 0.0002083658237718614, "epoch": 0.61, "step": 16560}
+{"loss": 0.7187, "learning_rate": 0.0002083101258772418, "epoch": 0.61, "step": 16570}
+{"loss": 0.5357, "learning_rate": 0.00020825442798262222, "epoch": 0.61, "step": 16580}
+{"loss": 0.5917, "learning_rate": 0.00020819873008800263, "epoch": 0.61, "step": 16590}
+{"loss": 0.6529, "learning_rate": 0.0002081430321933831, "epoch": 0.62, "step": 16600}
+{"loss": 0.6148, "learning_rate": 0.0002080873342987635, "epoch": 0.62, "step": 16610}
+{"loss": 0.7299, "learning_rate": 0.00020803163640414392, "epoch": 0.62, "step": 16620}
+{"loss": 0.575, "learning_rate": 0.00020797593850952433, "epoch": 0.62, "step": 16630}
+{"loss": 0.6312, "learning_rate": 0.00020792024061490474, "epoch": 0.62, "step": 16640}
+{"loss": 0.7628, "learning_rate": 0.00020786454272028515, "epoch": 0.62, "step": 16650}
+{"loss": 0.7716, "learning_rate": 0.0002078088448256656, "epoch": 0.62, "step": 16660}
+{"loss": 0.7695, "learning_rate": 0.000207753146931046, "epoch": 0.62, "step": 16670}
+{"loss": 0.5351, "learning_rate": 0.0002076974490364264, "epoch": 0.62, "step": 16680}
+{"loss": 0.7956, "learning_rate": 0.00020764175114180682, "epoch": 0.62, "step": 16690}
+{"loss": 0.7127, "learning_rate": 0.00020758605324718723, "epoch": 0.62, "step": 16700}
+{"loss": 0.6495, "learning_rate": 0.00020753035535256764, "epoch": 0.62, "step": 16710}
+{"loss": 0.6567, "learning_rate": 0.00020747465745794805, "epoch": 0.62, "step": 16720}
+{"loss": 0.5979, "learning_rate": 0.0002074189595633285, "epoch": 0.62, "step": 16730}
+{"loss": 0.6473, "learning_rate": 0.0002073632616687089, "epoch": 0.62, "step": 16740}
+{"loss": 0.7699, "learning_rate": 0.00020730756377408932, "epoch": 0.62, "step": 16750}
+{"loss": 0.6821, "learning_rate": 0.00020725186587946973, "epoch": 0.62, "step": 16760}
+{"loss": 0.6146, "learning_rate": 0.00020719616798485014, "epoch": 0.62, "step": 16770}
+{"loss": 0.6764, "learning_rate": 0.00020714047009023055, "epoch": 0.62, "step": 16780}
+{"loss": 0.688, "learning_rate": 0.00020708477219561101, "epoch": 0.62, "step": 16790}
+{"loss": 0.8138, "learning_rate": 0.00020702907430099142, "epoch": 0.62, "step": 16800}
+{"loss": 0.747, "learning_rate": 0.00020697337640637184, "epoch": 0.62, "step": 16810}
+{"loss": 0.4782, "learning_rate": 0.00020691767851175225, "epoch": 0.62, "step": 16820}
+{"loss": 0.5576, "learning_rate": 0.00020686198061713266, "epoch": 0.62, "step": 16830}
+{"loss": 0.6364, "learning_rate": 0.00020680628272251307, "epoch": 0.62, "step": 16840}
+{"loss": 0.6803, "learning_rate": 0.00020675058482789348, "epoch": 0.62, "step": 16850}
+{"loss": 0.7732, "learning_rate": 0.00020669488693327392, "epoch": 0.62, "step": 16860}
+{"loss": 0.7448, "learning_rate": 0.00020663918903865433, "epoch": 0.63, "step": 16870}
+{"loss": 0.6297, "learning_rate": 0.00020658349114403474, "epoch": 0.63, "step": 16880}
+{"loss": 0.6941, "learning_rate": 0.00020652779324941515, "epoch": 0.63, "step": 16890}
+{"loss": 0.6296, "learning_rate": 0.00020647209535479556, "epoch": 0.63, "step": 16900}
+{"loss": 0.609, "learning_rate": 0.00020641639746017597, "epoch": 0.63, "step": 16910}
+{"loss": 0.6334, "learning_rate": 0.0002063606995655564, "epoch": 0.63, "step": 16920}
+{"loss": 0.6894, "learning_rate": 0.00020630500167093682, "epoch": 0.63, "step": 16930}
+{"loss": 0.666, "learning_rate": 0.00020624930377631724, "epoch": 0.63, "step": 16940}
+{"loss": 0.6468, "learning_rate": 0.00020619360588169765, "epoch": 0.63, "step": 16950}
+{"loss": 0.5107, "learning_rate": 0.00020613790798707806, "epoch": 0.63, "step": 16960}
+{"loss": 0.698, "learning_rate": 0.00020608221009245847, "epoch": 0.63, "step": 16970}
+{"loss": 0.7222, "learning_rate": 0.00020602651219783888, "epoch": 0.63, "step": 16980}
+{"loss": 0.6228, "learning_rate": 0.00020597081430321934, "epoch": 0.63, "step": 16990}
+{"loss": 0.604, "learning_rate": 0.00020591511640859976, "epoch": 0.63, "step": 17000}
+{"loss": 0.6368, "learning_rate": 0.00020585941851398017, "epoch": 0.63, "step": 17010}
+{"loss": 0.7241, "learning_rate": 0.00020580372061936058, "epoch": 0.63, "step": 17020}
+{"loss": 0.66, "learning_rate": 0.000205748022724741, "epoch": 0.63, "step": 17030}
+{"loss": 0.662, "learning_rate": 0.0002056923248301214, "epoch": 0.63, "step": 17040}
+{"loss": 0.6612, "learning_rate": 0.00020563662693550184, "epoch": 0.63, "step": 17050}
+{"loss": 0.7032, "learning_rate": 0.00020558092904088225, "epoch": 0.63, "step": 17060}
+{"loss": 0.6303, "learning_rate": 0.00020552523114626266, "epoch": 0.63, "step": 17070}
+{"loss": 0.6672, "learning_rate": 0.00020546953325164307, "epoch": 0.63, "step": 17080}
+{"loss": 0.5975, "learning_rate": 0.00020541383535702348, "epoch": 0.63, "step": 17090}
+{"loss": 0.6077, "learning_rate": 0.0002053581374624039, "epoch": 0.63, "step": 17100}
+{"loss": 0.6935, "learning_rate": 0.0002053024395677843, "epoch": 0.63, "step": 17110}
+{"loss": 0.5895, "learning_rate": 0.00020524674167316474, "epoch": 0.63, "step": 17120}
+{"loss": 0.5868, "learning_rate": 0.00020519104377854516, "epoch": 0.63, "step": 17130}
+{"loss": 0.5879, "learning_rate": 0.00020513534588392557, "epoch": 0.64, "step": 17140}
+{"loss": 0.5866, "learning_rate": 0.00020507964798930598, "epoch": 0.64, "step": 17150}
+{"loss": 0.6544, "learning_rate": 0.0002050239500946864, "epoch": 0.64, "step": 17160}
+{"loss": 0.6674, "learning_rate": 0.0002049682522000668, "epoch": 0.64, "step": 17170}
+{"loss": 0.5829, "learning_rate": 0.00020491255430544726, "epoch": 0.64, "step": 17180}
+{"loss": 0.7002, "learning_rate": 0.00020485685641082768, "epoch": 0.64, "step": 17190}
+{"loss": 0.6267, "learning_rate": 0.0002048011585162081, "epoch": 0.64, "step": 17200}
+{"loss": 0.6676, "learning_rate": 0.0002047454606215885, "epoch": 0.64, "step": 17210}
+{"loss": 0.6414, "learning_rate": 0.0002046897627269689, "epoch": 0.64, "step": 17220}
+{"loss": 0.6435, "learning_rate": 0.00020463406483234932, "epoch": 0.64, "step": 17230}
+{"loss": 0.6674, "learning_rate": 0.00020457836693772973, "epoch": 0.64, "step": 17240}
+{"loss": 0.7893, "learning_rate": 0.00020452266904311017, "epoch": 0.64, "step": 17250}
+{"loss": 0.5701, "learning_rate": 0.00020446697114849058, "epoch": 0.64, "step": 17260}
+{"loss": 0.537, "learning_rate": 0.000204411273253871, "epoch": 0.64, "step": 17270}
+{"loss": 0.4973, "learning_rate": 0.0002043555753592514, "epoch": 0.64, "step": 17280}
+{"loss": 0.7679, "learning_rate": 0.00020429987746463181, "epoch": 0.64, "step": 17290}
+{"loss": 0.6712, "learning_rate": 0.00020424417957001223, "epoch": 0.64, "step": 17300}
+{"loss": 0.7045, "learning_rate": 0.00020418848167539266, "epoch": 0.64, "step": 17310}
+{"loss": 0.6852, "learning_rate": 0.00020413278378077308, "epoch": 0.64, "step": 17320}
+{"loss": 0.6592, "learning_rate": 0.00020407708588615349, "epoch": 0.64, "step": 17330}
+{"loss": 0.6403, "learning_rate": 0.0002040213879915339, "epoch": 0.64, "step": 17340}
+{"loss": 0.7452, "learning_rate": 0.0002039656900969143, "epoch": 0.64, "step": 17350}
+{"loss": 0.6997, "learning_rate": 0.00020390999220229472, "epoch": 0.64, "step": 17360}
+{"loss": 0.6126, "learning_rate": 0.00020385429430767513, "epoch": 0.64, "step": 17370}
+{"loss": 0.7993, "learning_rate": 0.00020379859641305557, "epoch": 0.64, "step": 17380}
+{"loss": 0.7044, "learning_rate": 0.00020374289851843598, "epoch": 0.64, "step": 17390}
+{"loss": 0.609, "learning_rate": 0.0002036872006238164, "epoch": 0.64, "step": 17400}
+{"loss": 0.643, "learning_rate": 0.00020363150272919683, "epoch": 0.65, "step": 17410}
+{"loss": 0.7644, "learning_rate": 0.00020357580483457724, "epoch": 0.65, "step": 17420}
+{"loss": 0.7087, "learning_rate": 0.00020352010693995765, "epoch": 0.65, "step": 17430}
+{"loss": 0.6256, "learning_rate": 0.0002034644090453381, "epoch": 0.65, "step": 17440}
+{"loss": 0.5963, "learning_rate": 0.0002034087111507185, "epoch": 0.65, "step": 17450}
+{"loss": 0.6397, "learning_rate": 0.0002033530132560989, "epoch": 0.65, "step": 17460}
+{"loss": 0.682, "learning_rate": 0.00020329731536147932, "epoch": 0.65, "step": 17470}
+{"loss": 0.7407, "learning_rate": 0.00020324161746685973, "epoch": 0.65, "step": 17480}
+{"loss": 0.7645, "learning_rate": 0.00020318591957224015, "epoch": 0.65, "step": 17490}
+{"loss": 0.6624, "learning_rate": 0.00020313022167762056, "epoch": 0.65, "step": 17500}
+{"loss": 0.5608, "learning_rate": 0.000203074523783001, "epoch": 0.65, "step": 17510}
+{"loss": 0.7884, "learning_rate": 0.0002030188258883814, "epoch": 0.65, "step": 17520}
+{"loss": 0.6766, "learning_rate": 0.00020296312799376182, "epoch": 0.65, "step": 17530}
+{"loss": 0.6252, "learning_rate": 0.00020290743009914223, "epoch": 0.65, "step": 17540}
+{"loss": 0.7199, "learning_rate": 0.00020285173220452264, "epoch": 0.65, "step": 17550}
+{"loss": 0.7059, "learning_rate": 0.00020279603430990305, "epoch": 0.65, "step": 17560}
+{"loss": 0.6424, "learning_rate": 0.0002027403364152835, "epoch": 0.65, "step": 17570}
+{"loss": 0.7056, "learning_rate": 0.0002026846385206639, "epoch": 0.65, "step": 17580}
+{"loss": 0.5768, "learning_rate": 0.0002026289406260443, "epoch": 0.65, "step": 17590}
+{"loss": 0.5203, "learning_rate": 0.00020257324273142472, "epoch": 0.65, "step": 17600}
+{"loss": 0.6972, "learning_rate": 0.00020251754483680513, "epoch": 0.65, "step": 17610}
+{"loss": 0.652, "learning_rate": 0.00020246184694218554, "epoch": 0.65, "step": 17620}
+{"loss": 0.5759, "learning_rate": 0.00020240614904756596, "epoch": 0.65, "step": 17630}
+{"loss": 0.6558, "learning_rate": 0.00020235045115294642, "epoch": 0.65, "step": 17640}
+{"loss": 0.6341, "learning_rate": 0.00020229475325832683, "epoch": 0.65, "step": 17650}
+{"loss": 0.7283, "learning_rate": 0.00020223905536370724, "epoch": 0.65, "step": 17660}
+{"loss": 0.7642, "learning_rate": 0.00020218335746908765, "epoch": 0.65, "step": 17670}
+{"loss": 0.6286, "learning_rate": 0.00020212765957446807, "epoch": 0.66, "step": 17680}
+{"loss": 0.6005, "learning_rate": 0.00020207196167984848, "epoch": 0.66, "step": 17690}
+{"loss": 0.6777, "learning_rate": 0.00020201626378522892, "epoch": 0.66, "step": 17700}
+{"loss": 0.5649, "learning_rate": 0.00020196056589060933, "epoch": 0.66, "step": 17710}
+{"loss": 0.68, "learning_rate": 0.00020190486799598974, "epoch": 0.66, "step": 17720}
+{"loss": 0.6041, "learning_rate": 0.00020184917010137015, "epoch": 0.66, "step": 17730}
+{"loss": 0.6246, "learning_rate": 0.00020179347220675056, "epoch": 0.66, "step": 17740}
+{"loss": 0.7433, "learning_rate": 0.00020173777431213097, "epoch": 0.66, "step": 17750}
+{"loss": 0.5697, "learning_rate": 0.00020168207641751138, "epoch": 0.66, "step": 17760}
+{"loss": 0.5997, "learning_rate": 0.00020162637852289182, "epoch": 0.66, "step": 17770}
+{"loss": 0.6056, "learning_rate": 0.00020157068062827223, "epoch": 0.66, "step": 17780}
+{"loss": 0.7026, "learning_rate": 0.00020151498273365264, "epoch": 0.66, "step": 17790}
+{"loss": 0.6083, "learning_rate": 0.00020145928483903305, "epoch": 0.66, "step": 17800}
+{"loss": 0.7213, "learning_rate": 0.00020140358694441346, "epoch": 0.66, "step": 17810}
+{"loss": 0.629, "learning_rate": 0.00020134788904979388, "epoch": 0.66, "step": 17820}
+{"loss": 0.7491, "learning_rate": 0.00020129219115517434, "epoch": 0.66, "step": 17830}
+{"loss": 0.5696, "learning_rate": 0.00020123649326055475, "epoch": 0.66, "step": 17840}
+{"loss": 0.7119, "learning_rate": 0.00020118079536593516, "epoch": 0.66, "step": 17850}
+{"loss": 0.7255, "learning_rate": 0.00020112509747131557, "epoch": 0.66, "step": 17860}
+{"loss": 0.5669, "learning_rate": 0.00020106939957669599, "epoch": 0.66, "step": 17870}
+{"loss": 0.6848, "learning_rate": 0.0002010137016820764, "epoch": 0.66, "step": 17880}
+{"loss": 0.5886, "learning_rate": 0.0002009580037874568, "epoch": 0.66, "step": 17890}
+{"loss": 0.5726, "learning_rate": 0.00020090230589283725, "epoch": 0.66, "step": 17900}
+{"loss": 0.6728, "learning_rate": 0.00020084660799821766, "epoch": 0.66, "step": 17910}
+{"loss": 0.6245, "learning_rate": 0.00020079091010359807, "epoch": 0.66, "step": 17920}
+{"loss": 0.6743, "learning_rate": 0.00020073521220897848, "epoch": 0.66, "step": 17930}
+{"loss": 0.5713, "learning_rate": 0.0002006795143143589, "epoch": 0.66, "step": 17940}
+{"loss": 0.6733, "learning_rate": 0.0002006238164197393, "epoch": 0.67, "step": 17950}
+{"loss": 0.661, "learning_rate": 0.00020056811852511974, "epoch": 0.67, "step": 17960}
+{"loss": 0.6789, "learning_rate": 0.00020051242063050015, "epoch": 0.67, "step": 17970}
+{"loss": 0.5861, "learning_rate": 0.00020045672273588056, "epoch": 0.67, "step": 17980}
+{"loss": 0.7447, "learning_rate": 0.00020040102484126097, "epoch": 0.67, "step": 17990}
+{"loss": 0.7067, "learning_rate": 0.00020034532694664138, "epoch": 0.67, "step": 18000}
+{"loss": 0.6465, "learning_rate": 0.0002002896290520218, "epoch": 0.67, "step": 18010}
+{"loss": 0.6069, "learning_rate": 0.0002002339311574022, "epoch": 0.67, "step": 18020}
+{"loss": 0.7092, "learning_rate": 0.00020017823326278267, "epoch": 0.67, "step": 18030}
+{"loss": 0.6169, "learning_rate": 0.00020012253536816308, "epoch": 0.67, "step": 18040}
+{"loss": 0.5533, "learning_rate": 0.0002000668374735435, "epoch": 0.67, "step": 18050}
+{"loss": 0.8475, "learning_rate": 0.0002000111395789239, "epoch": 0.67, "step": 18060}
+{"loss": 0.6603, "learning_rate": 0.00019995544168430432, "epoch": 0.67, "step": 18070}
+{"loss": 0.6336, "learning_rate": 0.00019989974378968473, "epoch": 0.67, "step": 18080}
+{"loss": 0.5824, "learning_rate": 0.00019984404589506517, "epoch": 0.67, "step": 18090}
+{"loss": 0.6861, "learning_rate": 0.00019978834800044558, "epoch": 0.67, "step": 18100}
+{"loss": 0.5162, "learning_rate": 0.000199732650105826, "epoch": 0.67, "step": 18110}
+{"loss": 0.6371, "learning_rate": 0.0001996769522112064, "epoch": 0.67, "step": 18120}
+{"loss": 0.6672, "learning_rate": 0.0001996212543165868, "epoch": 0.67, "step": 18130}
+{"loss": 0.6438, "learning_rate": 0.00019956555642196722, "epoch": 0.67, "step": 18140}
+{"loss": 0.6769, "learning_rate": 0.00019950985852734763, "epoch": 0.67, "step": 18150}
+{"loss": 0.8908, "learning_rate": 0.00019945416063272807, "epoch": 0.67, "step": 18160}
+{"loss": 0.6725, "learning_rate": 0.00019939846273810848, "epoch": 0.67, "step": 18170}
+{"loss": 0.6551, "learning_rate": 0.0001993427648434889, "epoch": 0.67, "step": 18180}
+{"loss": 0.6015, "learning_rate": 0.0001992870669488693, "epoch": 0.67, "step": 18190}
+{"loss": 0.5302, "learning_rate": 0.00019923136905424972, "epoch": 0.67, "step": 18200}
+{"loss": 0.5743, "learning_rate": 0.00019917567115963013, "epoch": 0.67, "step": 18210}
+{"loss": 0.6065, "learning_rate": 0.0001991199732650106, "epoch": 0.68, "step": 18220}
+{"loss": 0.7092, "learning_rate": 0.000199064275370391, "epoch": 0.68, "step": 18230}
+{"loss": 0.7828, "learning_rate": 0.00019900857747577141, "epoch": 0.68, "step": 18240}
+{"loss": 0.6355, "learning_rate": 0.00019895287958115183, "epoch": 0.68, "step": 18250}
+{"loss": 0.6551, "learning_rate": 0.00019889718168653224, "epoch": 0.68, "step": 18260}
+{"loss": 0.5744, "learning_rate": 0.00019884148379191265, "epoch": 0.68, "step": 18270}
+{"loss": 0.653, "learning_rate": 0.00019878578589729306, "epoch": 0.68, "step": 18280}
+{"loss": 0.7121, "learning_rate": 0.0001987300880026735, "epoch": 0.68, "step": 18290}
+{"loss": 0.6119, "learning_rate": 0.0001986743901080539, "epoch": 0.68, "step": 18300}
+{"loss": 0.6687, "learning_rate": 0.00019861869221343432, "epoch": 0.68, "step": 18310}
+{"loss": 0.6533, "learning_rate": 0.00019856299431881473, "epoch": 0.68, "step": 18320}
+{"loss": 0.78, "learning_rate": 0.00019850729642419514, "epoch": 0.68, "step": 18330}
+{"loss": 0.6069, "learning_rate": 0.00019845159852957555, "epoch": 0.68, "step": 18340}
+{"loss": 0.7008, "learning_rate": 0.000198395900634956, "epoch": 0.68, "step": 18350}
+{"loss": 0.7084, "learning_rate": 0.0001983402027403364, "epoch": 0.68, "step": 18360}
+{"loss": 0.63, "learning_rate": 0.0001982845048457168, "epoch": 0.68, "step": 18370}
+{"loss": 0.4814, "learning_rate": 0.00019822880695109722, "epoch": 0.68, "step": 18380}
+{"loss": 0.7997, "learning_rate": 0.00019817310905647764, "epoch": 0.68, "step": 18390}
+{"loss": 0.7286, "learning_rate": 0.00019811741116185805, "epoch": 0.68, "step": 18400}
+{"loss": 0.6538, "learning_rate": 0.00019806171326723849, "epoch": 0.68, "step": 18410}
+{"loss": 0.582, "learning_rate": 0.0001980060153726189, "epoch": 0.68, "step": 18420}
+{"loss": 0.6606, "learning_rate": 0.0001979503174779993, "epoch": 0.68, "step": 18430}
+{"loss": 0.9235, "learning_rate": 0.00019789461958337972, "epoch": 0.68, "step": 18440}
+{"loss": 0.6075, "learning_rate": 0.00019783892168876016, "epoch": 0.68, "step": 18450}
+{"loss": 0.6675, "learning_rate": 0.00019778322379414057, "epoch": 0.68, "step": 18460}
+{"loss": 0.6643, "learning_rate": 0.00019772752589952098, "epoch": 0.68, "step": 18470}
+{"loss": 0.6605, "learning_rate": 0.00019767182800490142, "epoch": 0.68, "step": 18480}
+{"loss": 0.5987, "learning_rate": 0.00019761613011028183, "epoch": 0.69, "step": 18490}
+{"loss": 0.6143, "learning_rate": 0.00019756043221566224, "epoch": 0.69, "step": 18500}
+{"loss": 0.6291, "learning_rate": 0.00019750473432104265, "epoch": 0.69, "step": 18510}
+{"loss": 0.6206, "learning_rate": 0.00019744903642642306, "epoch": 0.69, "step": 18520}
+{"loss": 0.6683, "learning_rate": 0.00019739333853180347, "epoch": 0.69, "step": 18530}
+{"loss": 0.6322, "learning_rate": 0.0001973376406371839, "epoch": 0.69, "step": 18540}
+{"loss": 0.6859, "learning_rate": 0.00019728194274256432, "epoch": 0.69, "step": 18550}
+{"loss": 0.5951, "learning_rate": 0.00019722624484794473, "epoch": 0.69, "step": 18560}
+{"loss": 0.6626, "learning_rate": 0.00019717054695332514, "epoch": 0.69, "step": 18570}
+{"loss": 0.593, "learning_rate": 0.00019711484905870556, "epoch": 0.69, "step": 18580}
+{"loss": 0.7337, "learning_rate": 0.00019705915116408597, "epoch": 0.69, "step": 18590}
+{"loss": 0.6291, "learning_rate": 0.00019700345326946638, "epoch": 0.69, "step": 18600}
+{"loss": 0.7644, "learning_rate": 0.00019694775537484682, "epoch": 0.69, "step": 18610}
+{"loss": 0.6554, "learning_rate": 0.00019689205748022723, "epoch": 0.69, "step": 18620}
+{"loss": 0.6219, "learning_rate": 0.00019683635958560764, "epoch": 0.69, "step": 18630}
+{"loss": 0.621, "learning_rate": 0.00019678066169098805, "epoch": 0.69, "step": 18640}
+{"loss": 0.7215, "learning_rate": 0.00019672496379636846, "epoch": 0.69, "step": 18650}
+{"loss": 0.6493, "learning_rate": 0.00019666926590174887, "epoch": 0.69, "step": 18660}
+{"loss": 0.625, "learning_rate": 0.00019661356800712934, "epoch": 0.69, "step": 18670}
+{"loss": 0.6352, "learning_rate": 0.00019655787011250975, "epoch": 0.69, "step": 18680}
+{"loss": 0.7213, "learning_rate": 0.00019650217221789016, "epoch": 0.69, "step": 18690}
+{"loss": 0.7981, "learning_rate": 0.00019644647432327057, "epoch": 0.69, "step": 18700}
+{"loss": 0.5422, "learning_rate": 0.00019639077642865098, "epoch": 0.69, "step": 18710}
+{"loss": 0.6416, "learning_rate": 0.0001963350785340314, "epoch": 0.69, "step": 18720}
+{"loss": 0.5892, "learning_rate": 0.0001962793806394118, "epoch": 0.69, "step": 18730}
+{"loss": 0.7087, "learning_rate": 0.00019622368274479224, "epoch": 0.69, "step": 18740}
+{"loss": 0.7, "learning_rate": 0.00019616798485017265, "epoch": 0.69, "step": 18750}
+{"loss": 0.6006, "learning_rate": 0.00019611228695555306, "epoch": 0.7, "step": 18760}
+{"loss": 0.5555, "learning_rate": 0.00019605658906093348, "epoch": 0.7, "step": 18770}
+{"loss": 0.7352, "learning_rate": 0.0001960008911663139, "epoch": 0.7, "step": 18780}
+{"loss": 0.6628, "learning_rate": 0.0001959451932716943, "epoch": 0.7, "step": 18790}
+{"loss": 0.7434, "learning_rate": 0.00019588949537707474, "epoch": 0.7, "step": 18800}
+{"loss": 0.7914, "learning_rate": 0.00019583379748245515, "epoch": 0.7, "step": 18810}
+{"loss": 0.6911, "learning_rate": 0.00019577809958783556, "epoch": 0.7, "step": 18820}
+{"loss": 0.6312, "learning_rate": 0.00019572240169321597, "epoch": 0.7, "step": 18830}
+{"loss": 0.6617, "learning_rate": 0.00019566670379859638, "epoch": 0.7, "step": 18840}
+{"loss": 0.8041, "learning_rate": 0.0001956110059039768, "epoch": 0.7, "step": 18850}
+{"loss": 0.5595, "learning_rate": 0.0001955553080093572, "epoch": 0.7, "step": 18860}
+{"loss": 0.6595, "learning_rate": 0.00019549961011473767, "epoch": 0.7, "step": 18870}
+{"loss": 0.6615, "learning_rate": 0.00019544391222011808, "epoch": 0.7, "step": 18880}
+{"loss": 0.5824, "learning_rate": 0.0001953882143254985, "epoch": 0.7, "step": 18890}
+{"loss": 0.847, "learning_rate": 0.0001953325164308789, "epoch": 0.7, "step": 18900}
+{"loss": 0.7381, "learning_rate": 0.0001952768185362593, "epoch": 0.7, "step": 18910}
+{"loss": 0.5843, "learning_rate": 0.00019522112064163972, "epoch": 0.7, "step": 18920}
+{"loss": 0.5933, "learning_rate": 0.00019516542274702016, "epoch": 0.7, "step": 18930}
+{"loss": 0.531, "learning_rate": 0.00019510972485240057, "epoch": 0.7, "step": 18940}
+{"loss": 0.6608, "learning_rate": 0.00019505402695778098, "epoch": 0.7, "step": 18950}
+{"loss": 0.5853, "learning_rate": 0.0001949983290631614, "epoch": 0.7, "step": 18960}
+{"loss": 0.6104, "learning_rate": 0.0001949426311685418, "epoch": 0.7, "step": 18970}
+{"loss": 0.586, "learning_rate": 0.00019488693327392222, "epoch": 0.7, "step": 18980}
+{"loss": 0.8872, "learning_rate": 0.00019483123537930263, "epoch": 0.7, "step": 18990}
+{"loss": 0.5051, "learning_rate": 0.00019477553748468307, "epoch": 0.7, "step": 19000}
+{"loss": 0.583, "learning_rate": 0.00019471983959006348, "epoch": 0.7, "step": 19010}
+{"loss": 0.627, "learning_rate": 0.0001946641416954439, "epoch": 0.7, "step": 19020}
+{"loss": 0.6354, "learning_rate": 0.0001946084438008243, "epoch": 0.71, "step": 19030}
+{"loss": 0.6424, "learning_rate": 0.0001945527459062047, "epoch": 0.71, "step": 19040}
+{"loss": 0.6479, "learning_rate": 0.00019449704801158512, "epoch": 0.71, "step": 19050}
+{"loss": 0.6363, "learning_rate": 0.0001944413501169656, "epoch": 0.71, "step": 19060}
+{"loss": 0.5707, "learning_rate": 0.000194385652222346, "epoch": 0.71, "step": 19070}
+{"loss": 0.6264, "learning_rate": 0.0001943299543277264, "epoch": 0.71, "step": 19080}
+{"loss": 0.741, "learning_rate": 0.00019427425643310682, "epoch": 0.71, "step": 19090}
+{"loss": 0.7087, "learning_rate": 0.00019421855853848723, "epoch": 0.71, "step": 19100}
+{"loss": 0.6304, "learning_rate": 0.00019416286064386764, "epoch": 0.71, "step": 19110}
+{"loss": 0.6409, "learning_rate": 0.00019410716274924806, "epoch": 0.71, "step": 19120}
+{"loss": 0.5855, "learning_rate": 0.0001940514648546285, "epoch": 0.71, "step": 19130}
+{"loss": 0.5732, "learning_rate": 0.0001939957669600089, "epoch": 0.71, "step": 19140}
+{"loss": 0.5669, "learning_rate": 0.00019394006906538932, "epoch": 0.71, "step": 19150}
+{"loss": 0.6845, "learning_rate": 0.00019388437117076973, "epoch": 0.71, "step": 19160}
+{"loss": 0.6808, "learning_rate": 0.00019382867327615014, "epoch": 0.71, "step": 19170}
+{"loss": 0.7629, "learning_rate": 0.00019377297538153055, "epoch": 0.71, "step": 19180}
+{"loss": 0.6542, "learning_rate": 0.000193717277486911, "epoch": 0.71, "step": 19190}
+{"loss": 0.7577, "learning_rate": 0.0001936615795922914, "epoch": 0.71, "step": 19200}
+{"loss": 0.5944, "learning_rate": 0.0001936058816976718, "epoch": 0.71, "step": 19210}
+{"loss": 0.7429, "learning_rate": 0.00019355018380305222, "epoch": 0.71, "step": 19220}
+{"loss": 0.604, "learning_rate": 0.00019349448590843263, "epoch": 0.71, "step": 19230}
+{"loss": 0.5441, "learning_rate": 0.00019343878801381304, "epoch": 0.71, "step": 19240}
+{"loss": 0.6282, "learning_rate": 0.00019338309011919345, "epoch": 0.71, "step": 19250}
+{"loss": 0.651, "learning_rate": 0.00019332739222457392, "epoch": 0.71, "step": 19260}
+{"loss": 0.8442, "learning_rate": 0.00019327169432995433, "epoch": 0.71, "step": 19270}
+{"loss": 0.6684, "learning_rate": 0.00019321599643533474, "epoch": 0.71, "step": 19280}
+{"loss": 0.7902, "learning_rate": 0.00019316029854071515, "epoch": 0.71, "step": 19290}
+{"loss": 0.672, "learning_rate": 0.00019310460064609556, "epoch": 0.72, "step": 19300}
+{"loss": 0.6835, "learning_rate": 0.00019304890275147598, "epoch": 0.72, "step": 19310}
+{"loss": 0.5991, "learning_rate": 0.0001929932048568564, "epoch": 0.72, "step": 19320}
+{"loss": 0.6007, "learning_rate": 0.00019293750696223682, "epoch": 0.72, "step": 19330}
+{"loss": 0.7035, "learning_rate": 0.00019288180906761724, "epoch": 0.72, "step": 19340}
+{"loss": 0.7652, "learning_rate": 0.00019282611117299765, "epoch": 0.72, "step": 19350}
+{"loss": 0.6993, "learning_rate": 0.00019277041327837806, "epoch": 0.72, "step": 19360}
+{"loss": 0.6339, "learning_rate": 0.00019271471538375847, "epoch": 0.72, "step": 19370}
+{"loss": 0.6306, "learning_rate": 0.00019265901748913888, "epoch": 0.72, "step": 19380}
+{"loss": 0.7027, "learning_rate": 0.00019260331959451932, "epoch": 0.72, "step": 19390}
+{"loss": 0.6746, "learning_rate": 0.00019254762169989973, "epoch": 0.72, "step": 19400}
+{"loss": 0.6782, "learning_rate": 0.00019249192380528014, "epoch": 0.72, "step": 19410}
+{"loss": 0.6163, "learning_rate": 0.00019243622591066055, "epoch": 0.72, "step": 19420}
+{"loss": 0.6623, "learning_rate": 0.00019238052801604096, "epoch": 0.72, "step": 19430}
+{"loss": 0.6736, "learning_rate": 0.00019232483012142137, "epoch": 0.72, "step": 19440}
+{"loss": 0.7566, "learning_rate": 0.0001922691322268018, "epoch": 0.72, "step": 19450}
+{"loss": 0.6878, "learning_rate": 0.00019221343433218222, "epoch": 0.72, "step": 19460}
+{"loss": 0.605, "learning_rate": 0.00019215773643756263, "epoch": 0.72, "step": 19470}
+{"loss": 0.5416, "learning_rate": 0.00019210203854294307, "epoch": 0.72, "step": 19480}
+{"loss": 0.738, "learning_rate": 0.00019204634064832348, "epoch": 0.72, "step": 19490}
+{"loss": 0.6201, "learning_rate": 0.0001919906427537039, "epoch": 0.72, "step": 19500}
+{"loss": 0.7031, "learning_rate": 0.00019194051464854628, "epoch": 0.72, "step": 19510}
+{"loss": 0.5377, "learning_rate": 0.0001918848167539267, "epoch": 0.72, "step": 19520}
+{"loss": 0.755, "learning_rate": 0.0001918291188593071, "epoch": 0.72, "step": 19530}
+{"loss": 0.77, "learning_rate": 0.00019177342096468751, "epoch": 0.72, "step": 19540}
+{"loss": 0.7279, "learning_rate": 0.00019171772307006793, "epoch": 0.72, "step": 19550}
+{"loss": 0.7072, "learning_rate": 0.00019166202517544836, "epoch": 0.72, "step": 19560}
+{"loss": 0.6377, "learning_rate": 0.00019160632728082878, "epoch": 0.73, "step": 19570}
+{"loss": 0.5932, "learning_rate": 0.0001915506293862092, "epoch": 0.73, "step": 19580}
+{"loss": 0.6916, "learning_rate": 0.0001914949314915896, "epoch": 0.73, "step": 19590}
+{"loss": 0.6103, "learning_rate": 0.00019143923359697, "epoch": 0.73, "step": 19600}
+{"loss": 0.7542, "learning_rate": 0.00019138353570235042, "epoch": 0.73, "step": 19610}
+{"loss": 0.6895, "learning_rate": 0.00019132783780773086, "epoch": 0.73, "step": 19620}
+{"loss": 0.6957, "learning_rate": 0.00019127213991311127, "epoch": 0.73, "step": 19630}
+{"loss": 0.7112, "learning_rate": 0.00019121644201849168, "epoch": 0.73, "step": 19640}
+{"loss": 0.6982, "learning_rate": 0.0001911607441238721, "epoch": 0.73, "step": 19650}
+{"loss": 0.6008, "learning_rate": 0.0001911050462292525, "epoch": 0.73, "step": 19660}
+{"loss": 0.5489, "learning_rate": 0.00019104934833463291, "epoch": 0.73, "step": 19670}
+{"loss": 0.6644, "learning_rate": 0.00019099365044001333, "epoch": 0.73, "step": 19680}
+{"loss": 0.6523, "learning_rate": 0.0001909379525453938, "epoch": 0.73, "step": 19690}
+{"loss": 0.6632, "learning_rate": 0.0001908822546507742, "epoch": 0.73, "step": 19700}
+{"loss": 0.7229, "learning_rate": 0.0001908265567561546, "epoch": 0.73, "step": 19710}
+{"loss": 0.6739, "learning_rate": 0.00019077085886153502, "epoch": 0.73, "step": 19720}
+{"loss": 0.6678, "learning_rate": 0.00019071516096691543, "epoch": 0.73, "step": 19730}
+{"loss": 0.6929, "learning_rate": 0.00019065946307229585, "epoch": 0.73, "step": 19740}
+{"loss": 0.6331, "learning_rate": 0.00019060376517767628, "epoch": 0.73, "step": 19750}
+{"loss": 0.7414, "learning_rate": 0.0001905480672830567, "epoch": 0.73, "step": 19760}
+{"loss": 0.7257, "learning_rate": 0.0001904923693884371, "epoch": 0.73, "step": 19770}
+{"loss": 0.7176, "learning_rate": 0.00019043667149381752, "epoch": 0.73, "step": 19780}
+{"loss": 0.6459, "learning_rate": 0.00019038097359919793, "epoch": 0.73, "step": 19790}
+{"loss": 0.8041, "learning_rate": 0.00019032527570457834, "epoch": 0.73, "step": 19800}
+{"loss": 0.6608, "learning_rate": 0.00019026957780995875, "epoch": 0.73, "step": 19810}
+{"loss": 0.6923, "learning_rate": 0.0001902138799153392, "epoch": 0.73, "step": 19820}
+{"loss": 0.6175, "learning_rate": 0.0001901581820207196, "epoch": 0.73, "step": 19830}
+{"loss": 0.6667, "learning_rate": 0.0001901024841261, "epoch": 0.74, "step": 19840}
+{"loss": 0.7444, "learning_rate": 0.00019004678623148042, "epoch": 0.74, "step": 19850}
+{"loss": 0.6638, "learning_rate": 0.00018999108833686083, "epoch": 0.74, "step": 19860}
+{"loss": 0.6069, "learning_rate": 0.00018993539044224125, "epoch": 0.74, "step": 19870}
+{"loss": 0.6109, "learning_rate": 0.0001898796925476217, "epoch": 0.74, "step": 19880}
+{"loss": 0.6163, "learning_rate": 0.00018982399465300212, "epoch": 0.74, "step": 19890}
+{"loss": 0.6371, "learning_rate": 0.00018976829675838253, "epoch": 0.74, "step": 19900}
+{"loss": 0.7007, "learning_rate": 0.00018971259886376294, "epoch": 0.74, "step": 19910}
+{"loss": 0.6362, "learning_rate": 0.00018965690096914335, "epoch": 0.74, "step": 19920}
+{"loss": 0.6207, "learning_rate": 0.00018960120307452377, "epoch": 0.74, "step": 19930}
+{"loss": 0.7749, "learning_rate": 0.00018954550517990418, "epoch": 0.74, "step": 19940}
+{"loss": 0.6498, "learning_rate": 0.00018948980728528462, "epoch": 0.74, "step": 19950}
+{"loss": 0.605, "learning_rate": 0.00018943410939066503, "epoch": 0.74, "step": 19960}
+{"loss": 0.702, "learning_rate": 0.00018937841149604544, "epoch": 0.74, "step": 19970}
+{"loss": 0.5961, "learning_rate": 0.00018932271360142585, "epoch": 0.74, "step": 19980}
+{"loss": 0.6399, "learning_rate": 0.00018926701570680626, "epoch": 0.74, "step": 19990}
+{"loss": 0.5963, "learning_rate": 0.00018921131781218667, "epoch": 0.74, "step": 20000}
+{"loss": 0.6984, "learning_rate": 0.0001891556199175671, "epoch": 0.74, "step": 20010}
+{"loss": 0.6384, "learning_rate": 0.00018909992202294752, "epoch": 0.74, "step": 20020}
+{"loss": 0.6585, "learning_rate": 0.00018904422412832793, "epoch": 0.74, "step": 20030}
+{"loss": 0.6981, "learning_rate": 0.00018898852623370834, "epoch": 0.74, "step": 20040}
+{"loss": 0.7814, "learning_rate": 0.00018893282833908875, "epoch": 0.74, "step": 20050}
+{"loss": 0.6603, "learning_rate": 0.00018887713044446917, "epoch": 0.74, "step": 20060}
+{"loss": 0.7115, "learning_rate": 0.00018882143254984958, "epoch": 0.74, "step": 20070}
+{"loss": 0.6559, "learning_rate": 0.00018876573465523001, "epoch": 0.74, "step": 20080}
+{"loss": 0.6861, "learning_rate": 0.00018871003676061043, "epoch": 0.74, "step": 20090}
+{"loss": 0.6876, "learning_rate": 0.00018865433886599086, "epoch": 0.74, "step": 20100}
+{"loss": 0.5808, "learning_rate": 0.00018859864097137127, "epoch": 0.75, "step": 20110}
+{"loss": 0.6236, "learning_rate": 0.00018854294307675169, "epoch": 0.75, "step": 20120}
+{"loss": 0.6091, "learning_rate": 0.0001884872451821321, "epoch": 0.75, "step": 20130}
+{"loss": 0.655, "learning_rate": 0.00018843154728751254, "epoch": 0.75, "step": 20140}
+{"loss": 0.7106, "learning_rate": 0.00018837584939289295, "epoch": 0.75, "step": 20150}
+{"loss": 0.7765, "learning_rate": 0.00018832015149827336, "epoch": 0.75, "step": 20160}
+{"loss": 0.5648, "learning_rate": 0.00018826445360365377, "epoch": 0.75, "step": 20170}
+{"loss": 0.5975, "learning_rate": 0.00018820875570903418, "epoch": 0.75, "step": 20180}
+{"loss": 0.6609, "learning_rate": 0.0001881530578144146, "epoch": 0.75, "step": 20190}
+{"loss": 0.7302, "learning_rate": 0.000188097359919795, "epoch": 0.75, "step": 20200}
+{"loss": 0.7204, "learning_rate": 0.00018804166202517544, "epoch": 0.75, "step": 20210}
+{"loss": 0.8433, "learning_rate": 0.00018798596413055585, "epoch": 0.75, "step": 20220}
+{"loss": 0.7414, "learning_rate": 0.00018793026623593626, "epoch": 0.75, "step": 20230}
+{"loss": 0.6493, "learning_rate": 0.00018787456834131667, "epoch": 0.75, "step": 20240}
+{"loss": 0.7527, "learning_rate": 0.00018781887044669709, "epoch": 0.75, "step": 20250}
+{"loss": 0.5422, "learning_rate": 0.0001877631725520775, "epoch": 0.75, "step": 20260}
+{"loss": 0.699, "learning_rate": 0.00018770747465745793, "epoch": 0.75, "step": 20270}
+{"loss": 0.6425, "learning_rate": 0.00018765177676283835, "epoch": 0.75, "step": 20280}
+{"loss": 0.6404, "learning_rate": 0.00018759607886821876, "epoch": 0.75, "step": 20290}
+{"loss": 0.6666, "learning_rate": 0.00018754038097359917, "epoch": 0.75, "step": 20300}
+{"loss": 0.6949, "learning_rate": 0.00018748468307897958, "epoch": 0.75, "step": 20310}
+{"loss": 0.5536, "learning_rate": 0.00018742898518436, "epoch": 0.75, "step": 20320}
+{"loss": 0.6066, "learning_rate": 0.00018737328728974046, "epoch": 0.75, "step": 20330}
+{"loss": 0.6386, "learning_rate": 0.00018731758939512087, "epoch": 0.75, "step": 20340}
+{"loss": 0.5856, "learning_rate": 0.00018726189150050128, "epoch": 0.75, "step": 20350}
+{"loss": 0.6494, "learning_rate": 0.0001872061936058817, "epoch": 0.75, "step": 20360}
+{"loss": 0.7225, "learning_rate": 0.0001871504957112621, "epoch": 0.75, "step": 20370}
+{"loss": 0.6422, "learning_rate": 0.0001870947978166425, "epoch": 0.76, "step": 20380}
+{"loss": 0.6051, "learning_rate": 0.00018703909992202292, "epoch": 0.76, "step": 20390}
+{"loss": 0.7431, "learning_rate": 0.00018698340202740336, "epoch": 0.76, "step": 20400}
+{"loss": 0.57, "learning_rate": 0.00018692770413278377, "epoch": 0.76, "step": 20410}
+{"loss": 0.5635, "learning_rate": 0.00018687200623816418, "epoch": 0.76, "step": 20420}
+{"loss": 0.5906, "learning_rate": 0.0001868163083435446, "epoch": 0.76, "step": 20430}
+{"loss": 0.6621, "learning_rate": 0.000186760610448925, "epoch": 0.76, "step": 20440}
+{"loss": 0.6276, "learning_rate": 0.00018670491255430542, "epoch": 0.76, "step": 20450}
+{"loss": 0.6124, "learning_rate": 0.00018664921465968585, "epoch": 0.76, "step": 20460}
+{"loss": 0.6845, "learning_rate": 0.00018659351676506627, "epoch": 0.76, "step": 20470}
+{"loss": 0.7457, "learning_rate": 0.00018653781887044668, "epoch": 0.76, "step": 20480}
+{"loss": 0.7103, "learning_rate": 0.0001864821209758271, "epoch": 0.76, "step": 20490}
+{"loss": 0.7599, "learning_rate": 0.0001864264230812075, "epoch": 0.76, "step": 20500}
+{"loss": 0.6047, "learning_rate": 0.0001863707251865879, "epoch": 0.76, "step": 20510}
+{"loss": 0.7335, "learning_rate": 0.00018631502729196832, "epoch": 0.76, "step": 20520}
+{"loss": 0.6653, "learning_rate": 0.0001862593293973488, "epoch": 0.76, "step": 20530}
+{"loss": 0.7796, "learning_rate": 0.0001862036315027292, "epoch": 0.76, "step": 20540}
+{"loss": 0.7679, "learning_rate": 0.0001861479336081096, "epoch": 0.76, "step": 20550}
+{"loss": 0.5963, "learning_rate": 0.00018609223571349002, "epoch": 0.76, "step": 20560}
+{"loss": 0.6968, "learning_rate": 0.00018603653781887043, "epoch": 0.76, "step": 20570}
+{"loss": 0.6086, "learning_rate": 0.00018598083992425084, "epoch": 0.76, "step": 20580}
+{"loss": 0.6239, "learning_rate": 0.00018592514202963128, "epoch": 0.76, "step": 20590}
+{"loss": 0.6947, "learning_rate": 0.0001858694441350117, "epoch": 0.76, "step": 20600}
+{"loss": 0.6271, "learning_rate": 0.0001858137462403921, "epoch": 0.76, "step": 20610}
+{"loss": 0.7061, "learning_rate": 0.00018575804834577251, "epoch": 0.76, "step": 20620}
+{"loss": 0.6634, "learning_rate": 0.00018570235045115293, "epoch": 0.76, "step": 20630}
+{"loss": 0.6456, "learning_rate": 0.00018564665255653334, "epoch": 0.76, "step": 20640}
+{"loss": 0.8379, "learning_rate": 0.00018559095466191375, "epoch": 0.77, "step": 20650}
+{"loss": 0.6626, "learning_rate": 0.00018553525676729419, "epoch": 0.77, "step": 20660}
+{"loss": 0.6343, "learning_rate": 0.0001854795588726746, "epoch": 0.77, "step": 20670}
+{"loss": 0.6303, "learning_rate": 0.000185423860978055, "epoch": 0.77, "step": 20680}
+{"loss": 0.5683, "learning_rate": 0.00018536816308343542, "epoch": 0.77, "step": 20690}
+{"loss": 0.6122, "learning_rate": 0.00018531246518881583, "epoch": 0.77, "step": 20700}
+{"loss": 0.6551, "learning_rate": 0.00018525676729419624, "epoch": 0.77, "step": 20710}
+{"loss": 0.6795, "learning_rate": 0.0001852010693995767, "epoch": 0.77, "step": 20720}
+{"loss": 0.7068, "learning_rate": 0.00018514537150495712, "epoch": 0.77, "step": 20730}
+{"loss": 0.5266, "learning_rate": 0.00018508967361033753, "epoch": 0.77, "step": 20740}
+{"loss": 0.651, "learning_rate": 0.00018503397571571794, "epoch": 0.77, "step": 20750}
+{"loss": 0.6797, "learning_rate": 0.00018497827782109835, "epoch": 0.77, "step": 20760}
+{"loss": 0.6936, "learning_rate": 0.00018492257992647876, "epoch": 0.77, "step": 20770}
+{"loss": 0.6737, "learning_rate": 0.00018486688203185917, "epoch": 0.77, "step": 20780}
+{"loss": 0.7991, "learning_rate": 0.0001848111841372396, "epoch": 0.77, "step": 20790}
+{"loss": 0.6172, "learning_rate": 0.00018475548624262002, "epoch": 0.77, "step": 20800}
+{"loss": 0.676, "learning_rate": 0.00018469978834800043, "epoch": 0.77, "step": 20810}
+{"loss": 0.6675, "learning_rate": 0.00018464409045338085, "epoch": 0.77, "step": 20820}
+{"loss": 0.5879, "learning_rate": 0.00018458839255876126, "epoch": 0.77, "step": 20830}
+{"loss": 0.6728, "learning_rate": 0.00018453269466414167, "epoch": 0.77, "step": 20840}
+{"loss": 0.6116, "learning_rate": 0.0001844769967695221, "epoch": 0.77, "step": 20850}
+{"loss": 0.5568, "learning_rate": 0.00018442129887490252, "epoch": 0.77, "step": 20860}
+{"loss": 0.6798, "learning_rate": 0.00018436560098028293, "epoch": 0.77, "step": 20870}
+{"loss": 0.662, "learning_rate": 0.00018430990308566334, "epoch": 0.77, "step": 20880}
+{"loss": 0.6456, "learning_rate": 0.00018425420519104375, "epoch": 0.77, "step": 20890}
+{"loss": 0.8552, "learning_rate": 0.00018419850729642416, "epoch": 0.77, "step": 20900}
+{"loss": 0.7782, "learning_rate": 0.00018414280940180457, "epoch": 0.77, "step": 20910}
+{"loss": 0.6797, "learning_rate": 0.00018408711150718504, "epoch": 0.78, "step": 20920}
+{"loss": 0.6356, "learning_rate": 0.00018403141361256545, "epoch": 0.78, "step": 20930}
+{"loss": 0.6465, "learning_rate": 0.00018397571571794586, "epoch": 0.78, "step": 20940}
+{"loss": 0.6044, "learning_rate": 0.00018392001782332627, "epoch": 0.78, "step": 20950}
+{"loss": 0.5418, "learning_rate": 0.00018386431992870668, "epoch": 0.78, "step": 20960}
+{"loss": 0.6875, "learning_rate": 0.0001838086220340871, "epoch": 0.78, "step": 20970}
+{"loss": 0.7343, "learning_rate": 0.00018375292413946753, "epoch": 0.78, "step": 20980}
+{"loss": 0.5542, "learning_rate": 0.00018369722624484794, "epoch": 0.78, "step": 20990}
+{"loss": 0.6504, "learning_rate": 0.00018364152835022835, "epoch": 0.78, "step": 21000}
+{"loss": 0.6411, "learning_rate": 0.00018358583045560877, "epoch": 0.78, "step": 21010}
+{"loss": 0.6451, "learning_rate": 0.00018353013256098918, "epoch": 0.78, "step": 21020}
+{"loss": 0.6064, "learning_rate": 0.0001834744346663696, "epoch": 0.78, "step": 21030}
+{"loss": 0.6803, "learning_rate": 0.00018341873677175, "epoch": 0.78, "step": 21040}
+{"loss": 0.7642, "learning_rate": 0.00018336303887713044, "epoch": 0.78, "step": 21050}
+{"loss": 0.6216, "learning_rate": 0.00018330734098251085, "epoch": 0.78, "step": 21060}
+{"loss": 0.6669, "learning_rate": 0.00018325164308789126, "epoch": 0.78, "step": 21070}
+{"loss": 0.5003, "learning_rate": 0.00018319594519327167, "epoch": 0.78, "step": 21080}
+{"loss": 0.5968, "learning_rate": 0.00018314024729865208, "epoch": 0.78, "step": 21090}
+{"loss": 0.7301, "learning_rate": 0.0001830845494040325, "epoch": 0.78, "step": 21100}
+{"loss": 0.5535, "learning_rate": 0.00018302885150941293, "epoch": 0.78, "step": 21110}
+{"loss": 0.6121, "learning_rate": 0.00018297315361479334, "epoch": 0.78, "step": 21120}
+{"loss": 0.6282, "learning_rate": 0.00018291745572017378, "epoch": 0.78, "step": 21130}
+{"loss": 0.576, "learning_rate": 0.0001828617578255542, "epoch": 0.78, "step": 21140}
+{"loss": 0.7058, "learning_rate": 0.0001828060599309346, "epoch": 0.78, "step": 21150}
+{"loss": 0.6843, "learning_rate": 0.000182750362036315, "epoch": 0.78, "step": 21160}
+{"loss": 0.638, "learning_rate": 0.00018269466414169542, "epoch": 0.78, "step": 21170}
+{"loss": 0.5182, "learning_rate": 0.00018263896624707586, "epoch": 0.78, "step": 21180}
+{"loss": 0.6797, "learning_rate": 0.00018258326835245627, "epoch": 0.79, "step": 21190}
+{"loss": 0.752, "learning_rate": 0.00018252757045783669, "epoch": 0.79, "step": 21200}
+{"loss": 0.6137, "learning_rate": 0.0001824718725632171, "epoch": 0.79, "step": 21210}
+{"loss": 0.7475, "learning_rate": 0.0001824161746685975, "epoch": 0.79, "step": 21220}
+{"loss": 0.8034, "learning_rate": 0.00018236047677397792, "epoch": 0.79, "step": 21230}
+{"loss": 0.691, "learning_rate": 0.00018230477887935836, "epoch": 0.79, "step": 21240}
+{"loss": 0.5071, "learning_rate": 0.00018224908098473877, "epoch": 0.79, "step": 21250}
+{"loss": 0.6671, "learning_rate": 0.00018219338309011918, "epoch": 0.79, "step": 21260}
+{"loss": 0.7274, "learning_rate": 0.0001821376851954996, "epoch": 0.79, "step": 21270}
+{"loss": 0.7533, "learning_rate": 0.00018208198730088, "epoch": 0.79, "step": 21280}
+{"loss": 0.6973, "learning_rate": 0.0001820262894062604, "epoch": 0.79, "step": 21290}
+{"loss": 0.7688, "learning_rate": 0.00018197059151164082, "epoch": 0.79, "step": 21300}
+{"loss": 0.693, "learning_rate": 0.00018191489361702126, "epoch": 0.79, "step": 21310}
+{"loss": 0.6405, "learning_rate": 0.00018185919572240167, "epoch": 0.79, "step": 21320}
+{"loss": 0.6902, "learning_rate": 0.00018180349782778208, "epoch": 0.79, "step": 21330}
+{"loss": 0.661, "learning_rate": 0.0001817477999331625, "epoch": 0.79, "step": 21340}
+{"loss": 0.6779, "learning_rate": 0.0001816921020385429, "epoch": 0.79, "step": 21350}
+{"loss": 0.591, "learning_rate": 0.00018163640414392332, "epoch": 0.79, "step": 21360}
+{"loss": 0.8125, "learning_rate": 0.00018158070624930378, "epoch": 0.79, "step": 21370}
+{"loss": 0.6631, "learning_rate": 0.0001815250083546842, "epoch": 0.79, "step": 21380}
+{"loss": 0.6675, "learning_rate": 0.0001814693104600646, "epoch": 0.79, "step": 21390}
+{"loss": 0.7044, "learning_rate": 0.00018141361256544502, "epoch": 0.79, "step": 21400}
+{"loss": 0.6833, "learning_rate": 0.00018135791467082543, "epoch": 0.79, "step": 21410}
+{"loss": 0.626, "learning_rate": 0.00018130221677620584, "epoch": 0.79, "step": 21420}
+{"loss": 0.6701, "learning_rate": 0.00018124651888158625, "epoch": 0.79, "step": 21430}
+{"loss": 0.6441, "learning_rate": 0.0001811908209869667, "epoch": 0.79, "step": 21440}
+{"loss": 0.5859, "learning_rate": 0.0001811351230923471, "epoch": 0.8, "step": 21450}
+{"loss": 0.6164, "learning_rate": 0.0001810794251977275, "epoch": 0.8, "step": 21460}
+{"loss": 0.6975, "learning_rate": 0.00018102372730310792, "epoch": 0.8, "step": 21470}
+{"loss": 0.6003, "learning_rate": 0.00018096802940848833, "epoch": 0.8, "step": 21480}
+{"loss": 0.6407, "learning_rate": 0.00018091233151386874, "epoch": 0.8, "step": 21490}
+{"loss": 0.6791, "learning_rate": 0.00018085663361924918, "epoch": 0.8, "step": 21500}
+{"loss": 0.8427, "learning_rate": 0.00018080650551409154, "epoch": 0.8, "step": 21510}
+{"loss": 0.5963, "learning_rate": 0.00018075080761947195, "epoch": 0.8, "step": 21520}
+{"loss": 0.611, "learning_rate": 0.00018069510972485236, "epoch": 0.8, "step": 21530}
+{"loss": 0.6788, "learning_rate": 0.00018063941183023283, "epoch": 0.8, "step": 21540}
+{"loss": 0.6973, "learning_rate": 0.00018058371393561324, "epoch": 0.8, "step": 21550}
+{"loss": 0.6802, "learning_rate": 0.00018052801604099365, "epoch": 0.8, "step": 21560}
+{"loss": 0.7667, "learning_rate": 0.00018047231814637406, "epoch": 0.8, "step": 21570}
+{"loss": 0.6205, "learning_rate": 0.00018041662025175447, "epoch": 0.8, "step": 21580}
+{"loss": 0.6302, "learning_rate": 0.00018036092235713488, "epoch": 0.8, "step": 21590}
+{"loss": 0.5154, "learning_rate": 0.0001803052244625153, "epoch": 0.8, "step": 21600}
+{"loss": 0.7663, "learning_rate": 0.00018024952656789573, "epoch": 0.8, "step": 21610}
+{"loss": 0.6918, "learning_rate": 0.00018019382867327614, "epoch": 0.8, "step": 21620}
+{"loss": 0.8157, "learning_rate": 0.00018013813077865656, "epoch": 0.8, "step": 21630}
+{"loss": 0.761, "learning_rate": 0.00018008243288403697, "epoch": 0.8, "step": 21640}
+{"loss": 0.6386, "learning_rate": 0.00018002673498941738, "epoch": 0.8, "step": 21650}
+{"loss": 0.6069, "learning_rate": 0.0001799710370947978, "epoch": 0.8, "step": 21660}
+{"loss": 0.6493, "learning_rate": 0.00017991533920017823, "epoch": 0.8, "step": 21670}
+{"loss": 0.564, "learning_rate": 0.00017985964130555864, "epoch": 0.8, "step": 21680}
+{"loss": 0.5672, "learning_rate": 0.00017980394341093905, "epoch": 0.8, "step": 21690}
+{"loss": 0.5828, "learning_rate": 0.00017974824551631946, "epoch": 0.8, "step": 21700}
+{"loss": 0.6515, "learning_rate": 0.00017969254762169987, "epoch": 0.8, "step": 21710}
+{"loss": 0.5852, "learning_rate": 0.00017963684972708028, "epoch": 0.81, "step": 21720}
+{"loss": 0.7617, "learning_rate": 0.0001795811518324607, "epoch": 0.81, "step": 21730}
+{"loss": 0.7882, "learning_rate": 0.00017952545393784113, "epoch": 0.81, "step": 21740}
+{"loss": 0.7354, "learning_rate": 0.00017946975604322157, "epoch": 0.81, "step": 21750}
+{"loss": 0.7405, "learning_rate": 0.00017941405814860198, "epoch": 0.81, "step": 21760}
+{"loss": 0.6197, "learning_rate": 0.0001793583602539824, "epoch": 0.81, "step": 21770}
+{"loss": 0.7285, "learning_rate": 0.0001793026623593628, "epoch": 0.81, "step": 21780}
+{"loss": 0.6673, "learning_rate": 0.00017924696446474322, "epoch": 0.81, "step": 21790}
+{"loss": 0.6731, "learning_rate": 0.00017919126657012365, "epoch": 0.81, "step": 21800}
+{"loss": 0.6656, "learning_rate": 0.00017913556867550406, "epoch": 0.81, "step": 21810}
+{"loss": 0.6673, "learning_rate": 0.00017907987078088448, "epoch": 0.81, "step": 21820}
+{"loss": 0.7103, "learning_rate": 0.0001790241728862649, "epoch": 0.81, "step": 21830}
+{"loss": 0.6246, "learning_rate": 0.0001789684749916453, "epoch": 0.81, "step": 21840}
+{"loss": 0.7215, "learning_rate": 0.0001789127770970257, "epoch": 0.81, "step": 21850}
+{"loss": 0.6166, "learning_rate": 0.00017885707920240612, "epoch": 0.81, "step": 21860}
+{"loss": 0.6467, "learning_rate": 0.00017880138130778656, "epoch": 0.81, "step": 21870}
+{"loss": 0.6146, "learning_rate": 0.00017874568341316697, "epoch": 0.81, "step": 21880}
+{"loss": 0.6723, "learning_rate": 0.00017868998551854738, "epoch": 0.81, "step": 21890}
+{"loss": 0.6375, "learning_rate": 0.0001786342876239278, "epoch": 0.81, "step": 21900}
+{"loss": 0.6821, "learning_rate": 0.0001785785897293082, "epoch": 0.81, "step": 21910}
+{"loss": 0.6494, "learning_rate": 0.00017852289183468861, "epoch": 0.81, "step": 21920}
+{"loss": 0.6948, "learning_rate": 0.00017846719394006905, "epoch": 0.81, "step": 21930}
+{"loss": 0.5204, "learning_rate": 0.00017841149604544946, "epoch": 0.81, "step": 21940}
+{"loss": 0.6715, "learning_rate": 0.00017835579815082987, "epoch": 0.81, "step": 21950}
+{"loss": 0.6334, "learning_rate": 0.00017830010025621029, "epoch": 0.81, "step": 21960}
+{"loss": 0.6761, "learning_rate": 0.0001782444023615907, "epoch": 0.81, "step": 21970}
+{"loss": 0.6914, "learning_rate": 0.0001781887044669711, "epoch": 0.81, "step": 21980}
+{"loss": 0.6114, "learning_rate": 0.00017813300657235155, "epoch": 0.82, "step": 21990}
+{"loss": 0.684, "learning_rate": 0.00017807730867773198, "epoch": 0.82, "step": 22000}
+{"loss": 0.6062, "learning_rate": 0.0001780216107831124, "epoch": 0.82, "step": 22010}
+{"loss": 0.6965, "learning_rate": 0.0001779659128884928, "epoch": 0.82, "step": 22020}
+{"loss": 0.7334, "learning_rate": 0.00017791021499387322, "epoch": 0.82, "step": 22030}
+{"loss": 0.7175, "learning_rate": 0.00017785451709925363, "epoch": 0.82, "step": 22040}
+{"loss": 0.7129, "learning_rate": 0.00017779881920463404, "epoch": 0.82, "step": 22050}
+{"loss": 0.625, "learning_rate": 0.00017774312131001448, "epoch": 0.82, "step": 22060}
+{"loss": 0.5337, "learning_rate": 0.0001776874234153949, "epoch": 0.82, "step": 22070}
+{"loss": 0.7045, "learning_rate": 0.0001776317255207753, "epoch": 0.82, "step": 22080}
+{"loss": 0.61, "learning_rate": 0.0001775760276261557, "epoch": 0.82, "step": 22090}
+{"loss": 0.6923, "learning_rate": 0.00017752032973153612, "epoch": 0.82, "step": 22100}
+{"loss": 0.6564, "learning_rate": 0.00017746463183691653, "epoch": 0.82, "step": 22110}
+{"loss": 0.6906, "learning_rate": 0.00017740893394229695, "epoch": 0.82, "step": 22120}
+{"loss": 0.6841, "learning_rate": 0.00017735323604767738, "epoch": 0.82, "step": 22130}
+{"loss": 0.6599, "learning_rate": 0.0001772975381530578, "epoch": 0.82, "step": 22140}
+{"loss": 0.7184, "learning_rate": 0.0001772418402584382, "epoch": 0.82, "step": 22150}
+{"loss": 0.6033, "learning_rate": 0.00017718614236381862, "epoch": 0.82, "step": 22160}
+{"loss": 0.638, "learning_rate": 0.00017713044446919903, "epoch": 0.82, "step": 22170}
+{"loss": 0.6752, "learning_rate": 0.00017707474657457944, "epoch": 0.82, "step": 22180}
+{"loss": 0.6036, "learning_rate": 0.0001770190486799599, "epoch": 0.82, "step": 22190}
+{"loss": 0.6212, "learning_rate": 0.00017696335078534032, "epoch": 0.82, "step": 22200}
+{"loss": 0.6125, "learning_rate": 0.00017690765289072073, "epoch": 0.82, "step": 22210}
+{"loss": 0.7277, "learning_rate": 0.00017685195499610114, "epoch": 0.82, "step": 22220}
+{"loss": 0.6884, "learning_rate": 0.00017679625710148155, "epoch": 0.82, "step": 22230}
+{"loss": 0.6407, "learning_rate": 0.00017674055920686196, "epoch": 0.82, "step": 22240}
+{"loss": 0.5581, "learning_rate": 0.0001766848613122424, "epoch": 0.82, "step": 22250}
+{"loss": 0.6857, "learning_rate": 0.0001766291634176228, "epoch": 0.83, "step": 22260}
+{"loss": 0.6205, "learning_rate": 0.00017657346552300322, "epoch": 0.83, "step": 22270}
+{"loss": 0.7017, "learning_rate": 0.00017651776762838363, "epoch": 0.83, "step": 22280}
+{"loss": 0.6955, "learning_rate": 0.00017646206973376404, "epoch": 0.83, "step": 22290}
+{"loss": 0.7789, "learning_rate": 0.00017640637183914445, "epoch": 0.83, "step": 22300}
+{"loss": 0.7204, "learning_rate": 0.00017635067394452487, "epoch": 0.83, "step": 22310}
+{"loss": 0.6703, "learning_rate": 0.0001762949760499053, "epoch": 0.83, "step": 22320}
+{"loss": 0.6914, "learning_rate": 0.00017623927815528571, "epoch": 0.83, "step": 22330}
+{"loss": 0.702, "learning_rate": 0.00017618358026066613, "epoch": 0.83, "step": 22340}
+{"loss": 0.6555, "learning_rate": 0.00017612788236604654, "epoch": 0.83, "step": 22350}
+{"loss": 0.6718, "learning_rate": 0.00017607218447142695, "epoch": 0.83, "step": 22360}
+{"loss": 0.569, "learning_rate": 0.00017601648657680736, "epoch": 0.83, "step": 22370}
+{"loss": 0.6394, "learning_rate": 0.00017596078868218782, "epoch": 0.83, "step": 22380}
+{"loss": 0.8083, "learning_rate": 0.00017590509078756824, "epoch": 0.83, "step": 22390}
+{"loss": 0.5963, "learning_rate": 0.00017584939289294865, "epoch": 0.83, "step": 22400}
+{"loss": 0.6576, "learning_rate": 0.00017579369499832906, "epoch": 0.83, "step": 22410}
+{"loss": 0.5232, "learning_rate": 0.00017573799710370947, "epoch": 0.83, "step": 22420}
+{"loss": 0.6411, "learning_rate": 0.00017568229920908988, "epoch": 0.83, "step": 22430}
+{"loss": 0.6751, "learning_rate": 0.0001756266013144703, "epoch": 0.83, "step": 22440}
+{"loss": 0.6556, "learning_rate": 0.00017557090341985073, "epoch": 0.83, "step": 22450}
+{"loss": 0.6646, "learning_rate": 0.00017551520552523114, "epoch": 0.83, "step": 22460}
+{"loss": 0.7813, "learning_rate": 0.00017545950763061155, "epoch": 0.83, "step": 22470}
+{"loss": 0.5784, "learning_rate": 0.00017540380973599196, "epoch": 0.83, "step": 22480}
+{"loss": 0.6503, "learning_rate": 0.00017534811184137237, "epoch": 0.83, "step": 22490}
+{"loss": 0.5332, "learning_rate": 0.00017529241394675279, "epoch": 0.83, "step": 22500}
+{"loss": 0.6944, "learning_rate": 0.00017523671605213322, "epoch": 0.83, "step": 22510}
+{"loss": 0.6788, "learning_rate": 0.00017518101815751363, "epoch": 0.83, "step": 22520}
+{"loss": 0.657, "learning_rate": 0.00017512532026289405, "epoch": 0.84, "step": 22530}
+{"loss": 0.686, "learning_rate": 0.00017506962236827446, "epoch": 0.84, "step": 22540}
+{"loss": 0.7483, "learning_rate": 0.00017501392447365487, "epoch": 0.84, "step": 22550}
+{"loss": 0.7148, "learning_rate": 0.00017495822657903528, "epoch": 0.84, "step": 22560}
+{"loss": 0.7035, "learning_rate": 0.0001749025286844157, "epoch": 0.84, "step": 22570}
+{"loss": 0.6277, "learning_rate": 0.00017484683078979616, "epoch": 0.84, "step": 22580}
+{"loss": 0.7663, "learning_rate": 0.00017479113289517657, "epoch": 0.84, "step": 22590}
+{"loss": 0.6972, "learning_rate": 0.00017473543500055698, "epoch": 0.84, "step": 22600}
+{"loss": 0.7526, "learning_rate": 0.0001746797371059374, "epoch": 0.84, "step": 22610}
+{"loss": 0.5441, "learning_rate": 0.0001746240392113178, "epoch": 0.84, "step": 22620}
+{"loss": 0.6928, "learning_rate": 0.0001745683413166982, "epoch": 0.84, "step": 22630}
+{"loss": 0.6568, "learning_rate": 0.00017451264342207865, "epoch": 0.84, "step": 22640}
+{"loss": 0.6503, "learning_rate": 0.00017445694552745906, "epoch": 0.84, "step": 22650}
+{"loss": 0.6712, "learning_rate": 0.00017440124763283947, "epoch": 0.84, "step": 22660}
+{"loss": 0.7726, "learning_rate": 0.00017434554973821988, "epoch": 0.84, "step": 22670}
+{"loss": 0.7783, "learning_rate": 0.0001742898518436003, "epoch": 0.84, "step": 22680}
+{"loss": 0.5867, "learning_rate": 0.0001742341539489807, "epoch": 0.84, "step": 22690}
+{"loss": 0.652, "learning_rate": 0.00017417845605436112, "epoch": 0.84, "step": 22700}
+{"loss": 0.6259, "learning_rate": 0.00017412275815974155, "epoch": 0.84, "step": 22710}
+{"loss": 0.6372, "learning_rate": 0.00017406706026512197, "epoch": 0.84, "step": 22720}
+{"loss": 0.7315, "learning_rate": 0.00017401136237050238, "epoch": 0.84, "step": 22730}
+{"loss": 0.7145, "learning_rate": 0.0001739556644758828, "epoch": 0.84, "step": 22740}
+{"loss": 0.692, "learning_rate": 0.0001738999665812632, "epoch": 0.84, "step": 22750}
+{"loss": 0.5947, "learning_rate": 0.0001738442686866436, "epoch": 0.84, "step": 22760}
+{"loss": 0.5777, "learning_rate": 0.00017378857079202405, "epoch": 0.84, "step": 22770}
+{"loss": 0.5751, "learning_rate": 0.0001737328728974045, "epoch": 0.84, "step": 22780}
+{"loss": 0.6465, "learning_rate": 0.0001736771750027849, "epoch": 0.84, "step": 22790}
+{"loss": 0.6318, "learning_rate": 0.0001736214771081653, "epoch": 0.85, "step": 22800}
+{"loss": 0.6551, "learning_rate": 0.00017356577921354572, "epoch": 0.85, "step": 22810}
+{"loss": 0.5258, "learning_rate": 0.00017351008131892613, "epoch": 0.85, "step": 22820}
+{"loss": 0.6216, "learning_rate": 0.00017345438342430654, "epoch": 0.85, "step": 22830}
+{"loss": 0.6446, "learning_rate": 0.00017339868552968698, "epoch": 0.85, "step": 22840}
+{"loss": 0.6706, "learning_rate": 0.0001733429876350674, "epoch": 0.85, "step": 22850}
+{"loss": 0.6963, "learning_rate": 0.0001732872897404478, "epoch": 0.85, "step": 22860}
+{"loss": 0.5914, "learning_rate": 0.00017323159184582821, "epoch": 0.85, "step": 22870}
+{"loss": 0.5853, "learning_rate": 0.00017317589395120863, "epoch": 0.85, "step": 22880}
+{"loss": 0.6259, "learning_rate": 0.00017312019605658904, "epoch": 0.85, "step": 22890}
+{"loss": 0.6524, "learning_rate": 0.00017306449816196947, "epoch": 0.85, "step": 22900}
+{"loss": 0.8211, "learning_rate": 0.00017300880026734989, "epoch": 0.85, "step": 22910}
+{"loss": 0.6335, "learning_rate": 0.0001729531023727303, "epoch": 0.85, "step": 22920}
+{"loss": 0.679, "learning_rate": 0.0001728974044781107, "epoch": 0.85, "step": 22930}
+{"loss": 0.6377, "learning_rate": 0.00017284170658349112, "epoch": 0.85, "step": 22940}
+{"loss": 0.6331, "learning_rate": 0.00017278600868887153, "epoch": 0.85, "step": 22950}
+{"loss": 0.6661, "learning_rate": 0.00017273031079425194, "epoch": 0.85, "step": 22960}
+{"loss": 0.6344, "learning_rate": 0.00017267461289963238, "epoch": 0.85, "step": 22970}
+{"loss": 0.522, "learning_rate": 0.0001726189150050128, "epoch": 0.85, "step": 22980}
+{"loss": 0.6078, "learning_rate": 0.0001725632171103932, "epoch": 0.85, "step": 22990}
+{"loss": 0.6613, "learning_rate": 0.0001725075192157736, "epoch": 0.85, "step": 23000}
+{"loss": 0.7214, "learning_rate": 0.00017245182132115402, "epoch": 0.85, "step": 23010}
+{"loss": 0.6207, "learning_rate": 0.00017239612342653446, "epoch": 0.85, "step": 23020}
+{"loss": 0.587, "learning_rate": 0.0001723404255319149, "epoch": 0.85, "step": 23030}
+{"loss": 0.6524, "learning_rate": 0.0001722847276372953, "epoch": 0.85, "step": 23040}
+{"loss": 0.582, "learning_rate": 0.00017222902974267572, "epoch": 0.85, "step": 23050}
+{"loss": 0.6086, "learning_rate": 0.00017217333184805613, "epoch": 0.85, "step": 23060}
+{"loss": 0.7461, "learning_rate": 0.00017211763395343655, "epoch": 0.86, "step": 23070}
+{"loss": 0.6574, "learning_rate": 0.00017206193605881696, "epoch": 0.86, "step": 23080}
+{"loss": 0.5291, "learning_rate": 0.00017200623816419737, "epoch": 0.86, "step": 23090}
+{"loss": 0.622, "learning_rate": 0.0001719505402695778, "epoch": 0.86, "step": 23100}
+{"loss": 0.7323, "learning_rate": 0.00017189484237495822, "epoch": 0.86, "step": 23110}
+{"loss": 0.7087, "learning_rate": 0.00017183914448033863, "epoch": 0.86, "step": 23120}
+{"loss": 0.8162, "learning_rate": 0.00017178344658571904, "epoch": 0.86, "step": 23130}
+{"loss": 0.6663, "learning_rate": 0.00017172774869109945, "epoch": 0.86, "step": 23140}
+{"loss": 0.6863, "learning_rate": 0.00017167205079647986, "epoch": 0.86, "step": 23150}
+{"loss": 0.7078, "learning_rate": 0.0001716163529018603, "epoch": 0.86, "step": 23160}
+{"loss": 0.5336, "learning_rate": 0.0001715606550072407, "epoch": 0.86, "step": 23170}
+{"loss": 0.7072, "learning_rate": 0.00017150495711262112, "epoch": 0.86, "step": 23180}
+{"loss": 0.6792, "learning_rate": 0.00017144925921800153, "epoch": 0.86, "step": 23190}
+{"loss": 0.5563, "learning_rate": 0.00017139356132338194, "epoch": 0.86, "step": 23200}
+{"loss": 0.7307, "learning_rate": 0.00017133786342876236, "epoch": 0.86, "step": 23210}
+{"loss": 0.5978, "learning_rate": 0.00017128216553414277, "epoch": 0.86, "step": 23220}
+{"loss": 0.7075, "learning_rate": 0.00017122646763952323, "epoch": 0.86, "step": 23230}
+{"loss": 0.6717, "learning_rate": 0.00017117076974490364, "epoch": 0.86, "step": 23240}
+{"loss": 0.783, "learning_rate": 0.00017111507185028405, "epoch": 0.86, "step": 23250}
+{"loss": 0.6898, "learning_rate": 0.00017105937395566447, "epoch": 0.86, "step": 23260}
+{"loss": 0.6023, "learning_rate": 0.00017100367606104488, "epoch": 0.86, "step": 23270}
+{"loss": 0.6497, "learning_rate": 0.0001709479781664253, "epoch": 0.86, "step": 23280}
+{"loss": 0.6195, "learning_rate": 0.00017089228027180573, "epoch": 0.86, "step": 23290}
+{"loss": 0.7395, "learning_rate": 0.00017083658237718614, "epoch": 0.86, "step": 23300}
+{"loss": 0.6426, "learning_rate": 0.00017078088448256655, "epoch": 0.86, "step": 23310}
+{"loss": 0.6224, "learning_rate": 0.00017072518658794696, "epoch": 0.86, "step": 23320}
+{"loss": 0.6004, "learning_rate": 0.00017066948869332737, "epoch": 0.86, "step": 23330}
+{"loss": 0.6231, "learning_rate": 0.00017061379079870778, "epoch": 0.87, "step": 23340}
+{"loss": 0.6916, "learning_rate": 0.0001705580929040882, "epoch": 0.87, "step": 23350}
+{"loss": 0.6701, "learning_rate": 0.00017050239500946863, "epoch": 0.87, "step": 23360}
+{"loss": 0.7606, "learning_rate": 0.00017044669711484904, "epoch": 0.87, "step": 23370}
+{"loss": 0.6962, "learning_rate": 0.00017039099922022945, "epoch": 0.87, "step": 23380}
+{"loss": 0.5405, "learning_rate": 0.00017033530132560986, "epoch": 0.87, "step": 23390}
+{"loss": 0.5927, "learning_rate": 0.00017027960343099028, "epoch": 0.87, "step": 23400}
+{"loss": 0.7452, "learning_rate": 0.0001702239055363707, "epoch": 0.87, "step": 23410}
+{"loss": 0.7691, "learning_rate": 0.00017016820764175115, "epoch": 0.87, "step": 23420}
+{"loss": 0.7378, "learning_rate": 0.00017011250974713156, "epoch": 0.87, "step": 23430}
+{"loss": 0.7659, "learning_rate": 0.00017005681185251197, "epoch": 0.87, "step": 23440}
+{"loss": 0.5904, "learning_rate": 0.00017000111395789239, "epoch": 0.87, "step": 23450}
+{"loss": 0.8135, "learning_rate": 0.0001699454160632728, "epoch": 0.87, "step": 23460}
+{"loss": 0.5632, "learning_rate": 0.0001698897181686532, "epoch": 0.87, "step": 23470}
+{"loss": 0.6855, "learning_rate": 0.00016983402027403362, "epoch": 0.87, "step": 23480}
+{"loss": 0.6673, "learning_rate": 0.00016977832237941406, "epoch": 0.87, "step": 23490}
+{"loss": 0.572, "learning_rate": 0.00016972262448479447, "epoch": 0.87, "step": 23500}
+{"loss": 0.5296, "learning_rate": 0.00016966692659017488, "epoch": 0.87, "step": 23510}
+{"loss": 0.5319, "learning_rate": 0.00016961679848501724, "epoch": 0.87, "step": 23520}
+{"loss": 0.7523, "learning_rate": 0.00016956110059039768, "epoch": 0.87, "step": 23530}
+{"loss": 0.685, "learning_rate": 0.0001695054026957781, "epoch": 0.87, "step": 23540}
+{"loss": 0.6187, "learning_rate": 0.0001694497048011585, "epoch": 0.87, "step": 23550}
+{"loss": 0.6582, "learning_rate": 0.0001693940069065389, "epoch": 0.87, "step": 23560}
+{"loss": 0.6512, "learning_rate": 0.00016933830901191932, "epoch": 0.87, "step": 23570}
+{"loss": 0.7045, "learning_rate": 0.00016928261111729973, "epoch": 0.87, "step": 23580}
+{"loss": 0.745, "learning_rate": 0.00016922691322268017, "epoch": 0.87, "step": 23590}
+{"loss": 0.6822, "learning_rate": 0.00016917121532806058, "epoch": 0.87, "step": 23600}
+{"loss": 0.5626, "learning_rate": 0.000169115517433441, "epoch": 0.88, "step": 23610}
+{"loss": 0.4967, "learning_rate": 0.0001690598195388214, "epoch": 0.88, "step": 23620}
+{"loss": 0.7235, "learning_rate": 0.00016900412164420182, "epoch": 0.88, "step": 23630}
+{"loss": 0.5501, "learning_rate": 0.00016894842374958225, "epoch": 0.88, "step": 23640}
+{"loss": 0.7744, "learning_rate": 0.00016889272585496266, "epoch": 0.88, "step": 23650}
+{"loss": 0.6494, "learning_rate": 0.0001688370279603431, "epoch": 0.88, "step": 23660}
+{"loss": 0.6036, "learning_rate": 0.00016878133006572351, "epoch": 0.88, "step": 23670}
+{"loss": 0.612, "learning_rate": 0.00016872563217110393, "epoch": 0.88, "step": 23680}
+{"loss": 0.7653, "learning_rate": 0.00016866993427648434, "epoch": 0.88, "step": 23690}
+{"loss": 0.7363, "learning_rate": 0.00016861423638186475, "epoch": 0.88, "step": 23700}
+{"loss": 0.5293, "learning_rate": 0.00016855853848724516, "epoch": 0.88, "step": 23710}
+{"loss": 0.5533, "learning_rate": 0.0001685028405926256, "epoch": 0.88, "step": 23720}
+{"loss": 0.7485, "learning_rate": 0.000168447142698006, "epoch": 0.88, "step": 23730}
+{"loss": 0.7341, "learning_rate": 0.00016839144480338642, "epoch": 0.88, "step": 23740}
+{"loss": 0.5985, "learning_rate": 0.00016833574690876683, "epoch": 0.88, "step": 23750}
+{"loss": 0.5604, "learning_rate": 0.00016828004901414724, "epoch": 0.88, "step": 23760}
+{"loss": 0.5703, "learning_rate": 0.00016822435111952765, "epoch": 0.88, "step": 23770}
+{"loss": 0.6245, "learning_rate": 0.00016816865322490806, "epoch": 0.88, "step": 23780}
+{"loss": 0.6768, "learning_rate": 0.0001681129553302885, "epoch": 0.88, "step": 23790}
+{"loss": 0.6663, "learning_rate": 0.0001680572574356689, "epoch": 0.88, "step": 23800}
+{"loss": 0.6341, "learning_rate": 0.00016800155954104932, "epoch": 0.88, "step": 23810}
+{"loss": 0.7778, "learning_rate": 0.00016794586164642974, "epoch": 0.88, "step": 23820}
+{"loss": 0.612, "learning_rate": 0.00016789016375181015, "epoch": 0.88, "step": 23830}
+{"loss": 0.652, "learning_rate": 0.00016783446585719056, "epoch": 0.88, "step": 23840}
+{"loss": 0.5799, "learning_rate": 0.00016777876796257102, "epoch": 0.88, "step": 23850}
+{"loss": 0.776, "learning_rate": 0.00016772307006795143, "epoch": 0.88, "step": 23860}
+{"loss": 0.7102, "learning_rate": 0.00016766737217333185, "epoch": 0.88, "step": 23870}
+{"loss": 0.619, "learning_rate": 0.00016761167427871226, "epoch": 0.89, "step": 23880}
+{"loss": 0.706, "learning_rate": 0.00016755597638409267, "epoch": 0.89, "step": 23890}
+{"loss": 0.5741, "learning_rate": 0.00016750027848947308, "epoch": 0.89, "step": 23900}
+{"loss": 0.6037, "learning_rate": 0.0001674445805948535, "epoch": 0.89, "step": 23910}
+{"loss": 0.6974, "learning_rate": 0.00016738888270023393, "epoch": 0.89, "step": 23920}
+{"loss": 0.6473, "learning_rate": 0.00016733318480561434, "epoch": 0.89, "step": 23930}
+{"loss": 0.6176, "learning_rate": 0.00016727748691099475, "epoch": 0.89, "step": 23940}
+{"loss": 0.7261, "learning_rate": 0.00016722178901637516, "epoch": 0.89, "step": 23950}
+{"loss": 0.6298, "learning_rate": 0.00016716609112175557, "epoch": 0.89, "step": 23960}
+{"loss": 0.685, "learning_rate": 0.00016711039322713598, "epoch": 0.89, "step": 23970}
+{"loss": 0.6892, "learning_rate": 0.00016705469533251642, "epoch": 0.89, "step": 23980}
+{"loss": 0.7442, "learning_rate": 0.00016699899743789683, "epoch": 0.89, "step": 23990}
+{"loss": 0.52, "learning_rate": 0.00016694329954327724, "epoch": 0.89, "step": 24000}
+{"loss": 0.6932, "learning_rate": 0.00016688760164865766, "epoch": 0.89, "step": 24010}
+{"loss": 0.6591, "learning_rate": 0.00016683190375403807, "epoch": 0.89, "step": 24020}
+{"loss": 0.6619, "learning_rate": 0.00016677620585941848, "epoch": 0.89, "step": 24030}
+{"loss": 0.5335, "learning_rate": 0.0001667205079647989, "epoch": 0.89, "step": 24040}
+{"loss": 0.6035, "learning_rate": 0.00016666481007017935, "epoch": 0.89, "step": 24050}
+{"loss": 0.656, "learning_rate": 0.00016660911217555977, "epoch": 0.89, "step": 24060}
+{"loss": 0.6138, "learning_rate": 0.00016655341428094018, "epoch": 0.89, "step": 24070}
+{"loss": 0.6546, "learning_rate": 0.0001664977163863206, "epoch": 0.89, "step": 24080}
+{"loss": 0.698, "learning_rate": 0.000166442018491701, "epoch": 0.89, "step": 24090}
+{"loss": 0.6264, "learning_rate": 0.0001663863205970814, "epoch": 0.89, "step": 24100}
+{"loss": 0.626, "learning_rate": 0.00016633062270246185, "epoch": 0.89, "step": 24110}
+{"loss": 0.6154, "learning_rate": 0.00016627492480784226, "epoch": 0.89, "step": 24120}
+{"loss": 0.6858, "learning_rate": 0.00016621922691322267, "epoch": 0.89, "step": 24130}
+{"loss": 0.5447, "learning_rate": 0.00016616352901860308, "epoch": 0.89, "step": 24140}
+{"loss": 0.7641, "learning_rate": 0.0001661078311239835, "epoch": 0.9, "step": 24150}
+{"loss": 0.8559, "learning_rate": 0.0001660521332293639, "epoch": 0.9, "step": 24160}
+{"loss": 0.5974, "learning_rate": 0.00016599643533474434, "epoch": 0.9, "step": 24170}
+{"loss": 0.6061, "learning_rate": 0.00016594073744012475, "epoch": 0.9, "step": 24180}
+{"loss": 0.646, "learning_rate": 0.00016588503954550516, "epoch": 0.9, "step": 24190}
+{"loss": 0.7893, "learning_rate": 0.00016582934165088558, "epoch": 0.9, "step": 24200}
+{"loss": 0.6778, "learning_rate": 0.000165773643756266, "epoch": 0.9, "step": 24210}
+{"loss": 0.6634, "learning_rate": 0.0001657179458616464, "epoch": 0.9, "step": 24220}
+{"loss": 0.6203, "learning_rate": 0.0001656622479670268, "epoch": 0.9, "step": 24230}
+{"loss": 0.542, "learning_rate": 0.00016560655007240727, "epoch": 0.9, "step": 24240}
+{"loss": 0.5076, "learning_rate": 0.00016555085217778769, "epoch": 0.9, "step": 24250}
+{"loss": 0.7691, "learning_rate": 0.0001654951542831681, "epoch": 0.9, "step": 24260}
+{"loss": 0.5768, "learning_rate": 0.0001654394563885485, "epoch": 0.9, "step": 24270}
+{"loss": 0.8354, "learning_rate": 0.00016538375849392892, "epoch": 0.9, "step": 24280}
+{"loss": 0.6833, "learning_rate": 0.00016532806059930933, "epoch": 0.9, "step": 24290}
+{"loss": 0.6914, "learning_rate": 0.00016527236270468977, "epoch": 0.9, "step": 24300}
+{"loss": 0.7365, "learning_rate": 0.00016521666481007018, "epoch": 0.9, "step": 24310}
+{"loss": 0.581, "learning_rate": 0.0001651609669154506, "epoch": 0.9, "step": 24320}
+{"loss": 0.52, "learning_rate": 0.000165105269020831, "epoch": 0.9, "step": 24330}
+{"loss": 0.7986, "learning_rate": 0.0001650495711262114, "epoch": 0.9, "step": 24340}
+{"loss": 0.6155, "learning_rate": 0.00016499387323159182, "epoch": 0.9, "step": 24350}
+{"loss": 0.8769, "learning_rate": 0.00016493817533697223, "epoch": 0.9, "step": 24360}
+{"loss": 0.5053, "learning_rate": 0.00016488247744235267, "epoch": 0.9, "step": 24370}
+{"loss": 0.6034, "learning_rate": 0.00016482677954773308, "epoch": 0.9, "step": 24380}
+{"loss": 0.6345, "learning_rate": 0.0001647710816531135, "epoch": 0.9, "step": 24390}
+{"loss": 0.4832, "learning_rate": 0.0001647153837584939, "epoch": 0.9, "step": 24400}
+{"loss": 0.634, "learning_rate": 0.00016465968586387432, "epoch": 0.9, "step": 24410}
+{"loss": 0.5884, "learning_rate": 0.00016460398796925473, "epoch": 0.91, "step": 24420}
+{"loss": 0.6038, "learning_rate": 0.0001645482900746352, "epoch": 0.91, "step": 24430}
+{"loss": 0.6318, "learning_rate": 0.0001644925921800156, "epoch": 0.91, "step": 24440}
+{"loss": 0.6705, "learning_rate": 0.00016443689428539602, "epoch": 0.91, "step": 24450}
+{"loss": 0.7112, "learning_rate": 0.00016438119639077643, "epoch": 0.91, "step": 24460}
+{"loss": 0.5746, "learning_rate": 0.00016432549849615684, "epoch": 0.91, "step": 24470}
+{"loss": 0.6422, "learning_rate": 0.00016426980060153725, "epoch": 0.91, "step": 24480}
+{"loss": 0.5374, "learning_rate": 0.00016421410270691766, "epoch": 0.91, "step": 24490}
+{"loss": 0.7395, "learning_rate": 0.0001641584048122981, "epoch": 0.91, "step": 24500}
+{"loss": 0.6743, "learning_rate": 0.0001641027069176785, "epoch": 0.91, "step": 24510}
+{"loss": 0.6367, "learning_rate": 0.00016404700902305892, "epoch": 0.91, "step": 24520}
+{"loss": 0.6077, "learning_rate": 0.00016399131112843933, "epoch": 0.91, "step": 24530}
+{"loss": 0.6965, "learning_rate": 0.00016393561323381974, "epoch": 0.91, "step": 24540}
+{"loss": 0.8226, "learning_rate": 0.00016387991533920015, "epoch": 0.91, "step": 24550}
+{"loss": 0.685, "learning_rate": 0.0001638242174445806, "epoch": 0.91, "step": 24560}
+{"loss": 0.6093, "learning_rate": 0.000163768519549961, "epoch": 0.91, "step": 24570}
+{"loss": 0.7029, "learning_rate": 0.00016371282165534142, "epoch": 0.91, "step": 24580}
+{"loss": 0.7415, "learning_rate": 0.00016365712376072183, "epoch": 0.91, "step": 24590}
+{"loss": 0.7158, "learning_rate": 0.00016360142586610224, "epoch": 0.91, "step": 24600}
+{"loss": 0.7229, "learning_rate": 0.00016354572797148265, "epoch": 0.91, "step": 24610}
+{"loss": 0.6929, "learning_rate": 0.00016349003007686306, "epoch": 0.91, "step": 24620}
+{"loss": 0.7338, "learning_rate": 0.0001634343321822435, "epoch": 0.91, "step": 24630}
+{"loss": 0.7115, "learning_rate": 0.0001633786342876239, "epoch": 0.91, "step": 24640}
+{"loss": 0.5188, "learning_rate": 0.00016332293639300432, "epoch": 0.91, "step": 24650}
+{"loss": 0.6324, "learning_rate": 0.00016326723849838473, "epoch": 0.91, "step": 24660}
+{"loss": 0.5535, "learning_rate": 0.00016321154060376517, "epoch": 0.91, "step": 24670}
+{"loss": 0.591, "learning_rate": 0.00016315584270914558, "epoch": 0.91, "step": 24680}
+{"loss": 0.6096, "learning_rate": 0.00016310014481452602, "epoch": 0.92, "step": 24690}
+{"loss": 0.6553, "learning_rate": 0.00016304444691990643, "epoch": 0.92, "step": 24700}
+{"loss": 0.6191, "learning_rate": 0.00016298874902528684, "epoch": 0.92, "step": 24710}
+{"loss": 0.5887, "learning_rate": 0.00016293305113066725, "epoch": 0.92, "step": 24720}
+{"loss": 0.672, "learning_rate": 0.00016287735323604766, "epoch": 0.92, "step": 24730}
+{"loss": 0.5524, "learning_rate": 0.00016282165534142807, "epoch": 0.92, "step": 24740}
+{"loss": 0.5515, "learning_rate": 0.00016276595744680849, "epoch": 0.92, "step": 24750}
+{"loss": 0.6581, "learning_rate": 0.00016271025955218892, "epoch": 0.92, "step": 24760}
+{"loss": 0.7206, "learning_rate": 0.00016265456165756934, "epoch": 0.92, "step": 24770}
+{"loss": 0.554, "learning_rate": 0.00016259886376294975, "epoch": 0.92, "step": 24780}
+{"loss": 0.7198, "learning_rate": 0.00016254316586833016, "epoch": 0.92, "step": 24790}
+{"loss": 0.6821, "learning_rate": 0.00016248746797371057, "epoch": 0.92, "step": 24800}
+{"loss": 0.7164, "learning_rate": 0.00016243177007909098, "epoch": 0.92, "step": 24810}
+{"loss": 0.6463, "learning_rate": 0.00016237607218447142, "epoch": 0.92, "step": 24820}
+{"loss": 0.6428, "learning_rate": 0.00016232037428985183, "epoch": 0.92, "step": 24830}
+{"loss": 0.5694, "learning_rate": 0.00016226467639523224, "epoch": 0.92, "step": 24840}
+{"loss": 0.5452, "learning_rate": 0.00016220897850061265, "epoch": 0.92, "step": 24850}
+{"loss": 0.6166, "learning_rate": 0.00016215328060599306, "epoch": 0.92, "step": 24860}
+{"loss": 0.5327, "learning_rate": 0.00016209758271137347, "epoch": 0.92, "step": 24870}
+{"loss": 0.6611, "learning_rate": 0.00016204188481675388, "epoch": 0.92, "step": 24880}
+{"loss": 0.5479, "learning_rate": 0.00016198618692213435, "epoch": 0.92, "step": 24890}
+{"loss": 0.5646, "learning_rate": 0.00016193048902751476, "epoch": 0.92, "step": 24900}
+{"loss": 0.6692, "learning_rate": 0.00016187479113289517, "epoch": 0.92, "step": 24910}
+{"loss": 0.7033, "learning_rate": 0.00016181909323827558, "epoch": 0.92, "step": 24920}
+{"loss": 0.7868, "learning_rate": 0.000161763395343656, "epoch": 0.92, "step": 24930}
+{"loss": 0.6833, "learning_rate": 0.0001617076974490364, "epoch": 0.92, "step": 24940}
+{"loss": 0.7623, "learning_rate": 0.00016165199955441684, "epoch": 0.92, "step": 24950}
+{"loss": 0.662, "learning_rate": 0.00016159630165979726, "epoch": 0.93, "step": 24960}
+{"loss": 0.5462, "learning_rate": 0.00016154060376517767, "epoch": 0.93, "step": 24970}
+{"loss": 0.717, "learning_rate": 0.00016148490587055808, "epoch": 0.93, "step": 24980}
+{"loss": 0.5175, "learning_rate": 0.0001614292079759385, "epoch": 0.93, "step": 24990}
+{"loss": 0.7492, "learning_rate": 0.0001613735100813189, "epoch": 0.93, "step": 25000}
+{"loss": 0.6851, "learning_rate": 0.0001613178121866993, "epoch": 0.93, "step": 25010}
+{"loss": 0.643, "learning_rate": 0.00016126211429207975, "epoch": 0.93, "step": 25020}
+{"loss": 0.7349, "learning_rate": 0.00016120641639746016, "epoch": 0.93, "step": 25030}
+{"loss": 0.6396, "learning_rate": 0.00016115071850284057, "epoch": 0.93, "step": 25040}
+{"loss": 0.6984, "learning_rate": 0.00016109502060822098, "epoch": 0.93, "step": 25050}
+{"loss": 0.5591, "learning_rate": 0.0001610393227136014, "epoch": 0.93, "step": 25060}
+{"loss": 0.6068, "learning_rate": 0.0001609836248189818, "epoch": 0.93, "step": 25070}
+{"loss": 0.7693, "learning_rate": 0.00016092792692436227, "epoch": 0.93, "step": 25080}
+{"loss": 0.7127, "learning_rate": 0.00016087222902974268, "epoch": 0.93, "step": 25090}
+{"loss": 0.7093, "learning_rate": 0.0001608165311351231, "epoch": 0.93, "step": 25100}
+{"loss": 0.6731, "learning_rate": 0.0001607608332405035, "epoch": 0.93, "step": 25110}
+{"loss": 0.678, "learning_rate": 0.00016070513534588391, "epoch": 0.93, "step": 25120}
+{"loss": 0.7756, "learning_rate": 0.00016064943745126433, "epoch": 0.93, "step": 25130}
+{"loss": 0.7264, "learning_rate": 0.00016059373955664474, "epoch": 0.93, "step": 25140}
+{"loss": 0.8083, "learning_rate": 0.00016053804166202518, "epoch": 0.93, "step": 25150}
+{"loss": 0.6405, "learning_rate": 0.0001604823437674056, "epoch": 0.93, "step": 25160}
+{"loss": 0.6652, "learning_rate": 0.000160426645872786, "epoch": 0.93, "step": 25170}
+{"loss": 0.6089, "learning_rate": 0.0001603709479781664, "epoch": 0.93, "step": 25180}
+{"loss": 0.5786, "learning_rate": 0.00016031525008354682, "epoch": 0.93, "step": 25190}
+{"loss": 0.6221, "learning_rate": 0.00016025955218892723, "epoch": 0.93, "step": 25200}
+{"loss": 0.6511, "learning_rate": 0.00016020385429430767, "epoch": 0.93, "step": 25210}
+{"loss": 0.6086, "learning_rate": 0.00016014815639968808, "epoch": 0.93, "step": 25220}
+{"loss": 0.584, "learning_rate": 0.0001600924585050685, "epoch": 0.94, "step": 25230}
+{"loss": 0.631, "learning_rate": 0.0001600367606104489, "epoch": 0.94, "step": 25240}
+{"loss": 0.6757, "learning_rate": 0.00015998106271582931, "epoch": 0.94, "step": 25250}
+{"loss": 0.6999, "learning_rate": 0.00015992536482120972, "epoch": 0.94, "step": 25260}
+{"loss": 0.5917, "learning_rate": 0.00015986966692659014, "epoch": 0.94, "step": 25270}
+{"loss": 0.5864, "learning_rate": 0.0001598139690319706, "epoch": 0.94, "step": 25280}
+{"loss": 0.6321, "learning_rate": 0.000159758271137351, "epoch": 0.94, "step": 25290}
+{"loss": 0.7112, "learning_rate": 0.00015970257324273142, "epoch": 0.94, "step": 25300}
+{"loss": 0.5841, "learning_rate": 0.00015964687534811183, "epoch": 0.94, "step": 25310}
+{"loss": 0.6637, "learning_rate": 0.00015959117745349225, "epoch": 0.94, "step": 25320}
+{"loss": 0.5794, "learning_rate": 0.00015953547955887266, "epoch": 0.94, "step": 25330}
+{"loss": 0.712, "learning_rate": 0.0001594797816642531, "epoch": 0.94, "step": 25340}
+{"loss": 0.5649, "learning_rate": 0.0001594240837696335, "epoch": 0.94, "step": 25350}
+{"loss": 0.6051, "learning_rate": 0.00015936838587501392, "epoch": 0.94, "step": 25360}
+{"loss": 0.6621, "learning_rate": 0.00015931268798039433, "epoch": 0.94, "step": 25370}
+{"loss": 0.6029, "learning_rate": 0.00015925699008577474, "epoch": 0.94, "step": 25380}
+{"loss": 0.6278, "learning_rate": 0.00015920129219115515, "epoch": 0.94, "step": 25390}
+{"loss": 0.5972, "learning_rate": 0.00015914559429653556, "epoch": 0.94, "step": 25400}
+{"loss": 0.5629, "learning_rate": 0.000159089896401916, "epoch": 0.94, "step": 25410}
+{"loss": 0.5566, "learning_rate": 0.0001590341985072964, "epoch": 0.94, "step": 25420}
+{"loss": 0.7261, "learning_rate": 0.00015897850061267682, "epoch": 0.94, "step": 25430}
+{"loss": 0.6248, "learning_rate": 0.00015892280271805723, "epoch": 0.94, "step": 25440}
+{"loss": 0.7223, "learning_rate": 0.00015886710482343764, "epoch": 0.94, "step": 25450}
+{"loss": 0.7283, "learning_rate": 0.00015881140692881806, "epoch": 0.94, "step": 25460}
+{"loss": 0.6315, "learning_rate": 0.00015875570903419852, "epoch": 0.94, "step": 25470}
+{"loss": 0.5885, "learning_rate": 0.00015870001113957893, "epoch": 0.94, "step": 25480}
+{"loss": 0.665, "learning_rate": 0.00015864431324495934, "epoch": 0.94, "step": 25490}
+{"loss": 0.4955, "learning_rate": 0.00015858861535033975, "epoch": 0.95, "step": 25500}
+{"loss": 0.6782, "learning_rate": 0.00015853291745572017, "epoch": 0.95, "step": 25510}
+{"loss": 0.5667, "learning_rate": 0.00015847721956110058, "epoch": 0.95, "step": 25520}
+{"loss": 0.6292, "learning_rate": 0.000158421521666481, "epoch": 0.95, "step": 25530}
+{"loss": 0.664, "learning_rate": 0.00015836582377186143, "epoch": 0.95, "step": 25540}
+{"loss": 0.6065, "learning_rate": 0.00015831012587724184, "epoch": 0.95, "step": 25550}
+{"loss": 0.723, "learning_rate": 0.00015825442798262225, "epoch": 0.95, "step": 25560}
+{"loss": 0.6907, "learning_rate": 0.00015819873008800266, "epoch": 0.95, "step": 25570}
+{"loss": 0.7137, "learning_rate": 0.00015814303219338307, "epoch": 0.95, "step": 25580}
+{"loss": 0.7402, "learning_rate": 0.00015808733429876348, "epoch": 0.95, "step": 25590}
+{"loss": 0.593, "learning_rate": 0.00015803163640414392, "epoch": 0.95, "step": 25600}
+{"loss": 0.6021, "learning_rate": 0.00015797593850952433, "epoch": 0.95, "step": 25610}
+{"loss": 0.6045, "learning_rate": 0.00015792024061490474, "epoch": 0.95, "step": 25620}
+{"loss": 0.6405, "learning_rate": 0.00015786454272028515, "epoch": 0.95, "step": 25630}
+{"loss": 0.7044, "learning_rate": 0.00015780884482566556, "epoch": 0.95, "step": 25640}
+{"loss": 0.5941, "learning_rate": 0.00015775314693104598, "epoch": 0.95, "step": 25650}
+{"loss": 0.5414, "learning_rate": 0.0001576974490364264, "epoch": 0.95, "step": 25660}
+{"loss": 0.648, "learning_rate": 0.00015764175114180683, "epoch": 0.95, "step": 25670}
+{"loss": 0.5308, "learning_rate": 0.00015758605324718724, "epoch": 0.95, "step": 25680}
+{"loss": 0.8367, "learning_rate": 0.0001575359251420296, "epoch": 0.95, "step": 25690}
+{"loss": 0.6401, "learning_rate": 0.00015748022724741, "epoch": 0.95, "step": 25700}
+{"loss": 0.6349, "learning_rate": 0.00015742452935279047, "epoch": 0.95, "step": 25710}
+{"loss": 0.8038, "learning_rate": 0.00015736883145817088, "epoch": 0.95, "step": 25720}
+{"loss": 0.541, "learning_rate": 0.0001573131335635513, "epoch": 0.95, "step": 25730}
+{"loss": 0.4583, "learning_rate": 0.0001572574356689317, "epoch": 0.95, "step": 25740}
+{"loss": 0.7132, "learning_rate": 0.00015720173777431212, "epoch": 0.95, "step": 25750}
+{"loss": 0.71, "learning_rate": 0.00015714603987969253, "epoch": 0.95, "step": 25760}
+{"loss": 0.7718, "learning_rate": 0.00015709034198507297, "epoch": 0.96, "step": 25770}
+{"loss": 0.7438, "learning_rate": 0.00015703464409045338, "epoch": 0.96, "step": 25780}
+{"loss": 0.6256, "learning_rate": 0.0001569789461958338, "epoch": 0.96, "step": 25790}
+{"loss": 0.7839, "learning_rate": 0.0001569232483012142, "epoch": 0.96, "step": 25800}
+{"loss": 0.6743, "learning_rate": 0.0001568675504065946, "epoch": 0.96, "step": 25810}
+{"loss": 0.5459, "learning_rate": 0.00015681185251197502, "epoch": 0.96, "step": 25820}
+{"loss": 0.6577, "learning_rate": 0.00015675615461735543, "epoch": 0.96, "step": 25830}
+{"loss": 0.7282, "learning_rate": 0.00015670045672273587, "epoch": 0.96, "step": 25840}
+{"loss": 0.6497, "learning_rate": 0.00015664475882811628, "epoch": 0.96, "step": 25850}
+{"loss": 0.5238, "learning_rate": 0.0001565890609334967, "epoch": 0.96, "step": 25860}
+{"loss": 0.7108, "learning_rate": 0.0001565333630388771, "epoch": 0.96, "step": 25870}
+{"loss": 0.6008, "learning_rate": 0.00015647766514425752, "epoch": 0.96, "step": 25880}
+{"loss": 0.6701, "learning_rate": 0.00015642196724963793, "epoch": 0.96, "step": 25890}
+{"loss": 0.6457, "learning_rate": 0.0001563662693550184, "epoch": 0.96, "step": 25900}
+{"loss": 0.5937, "learning_rate": 0.0001563105714603988, "epoch": 0.96, "step": 25910}
+{"loss": 0.6476, "learning_rate": 0.00015625487356577921, "epoch": 0.96, "step": 25920}
+{"loss": 0.7736, "learning_rate": 0.00015619917567115963, "epoch": 0.96, "step": 25930}
+{"loss": 0.5294, "learning_rate": 0.00015614347777654004, "epoch": 0.96, "step": 25940}
+{"loss": 0.7819, "learning_rate": 0.00015608777988192045, "epoch": 0.96, "step": 25950}
+{"loss": 0.7433, "learning_rate": 0.00015603208198730086, "epoch": 0.96, "step": 25960}
+{"loss": 0.541, "learning_rate": 0.0001559763840926813, "epoch": 0.96, "step": 25970}
+{"loss": 0.6122, "learning_rate": 0.0001559206861980617, "epoch": 0.96, "step": 25980}
+{"loss": 0.7083, "learning_rate": 0.00015586498830344212, "epoch": 0.96, "step": 25990}
+{"loss": 0.6321, "learning_rate": 0.00015580929040882253, "epoch": 0.96, "step": 26000}
+{"loss": 0.6895, "learning_rate": 0.00015575359251420294, "epoch": 0.96, "step": 26010}
+{"loss": 0.6287, "learning_rate": 0.00015569789461958335, "epoch": 0.96, "step": 26020}
+{"loss": 0.6067, "learning_rate": 0.0001556421967249638, "epoch": 0.96, "step": 26030}
+{"loss": 0.7166, "learning_rate": 0.0001555864988303442, "epoch": 0.97, "step": 26040}
+{"loss": 0.6167, "learning_rate": 0.0001555308009357246, "epoch": 0.97, "step": 26050}
+{"loss": 0.7379, "learning_rate": 0.00015547510304110502, "epoch": 0.97, "step": 26060}
+{"loss": 0.5657, "learning_rate": 0.00015541940514648544, "epoch": 0.97, "step": 26070}
+{"loss": 0.6784, "learning_rate": 0.00015536370725186585, "epoch": 0.97, "step": 26080}
+{"loss": 0.6877, "learning_rate": 0.0001553080093572463, "epoch": 0.97, "step": 26090}
+{"loss": 0.9396, "learning_rate": 0.00015525231146262672, "epoch": 0.97, "step": 26100}
+{"loss": 0.7001, "learning_rate": 0.00015519661356800713, "epoch": 0.97, "step": 26110}
+{"loss": 0.6222, "learning_rate": 0.00015514091567338755, "epoch": 0.97, "step": 26120}
+{"loss": 0.5846, "learning_rate": 0.00015508521777876796, "epoch": 0.97, "step": 26130}
+{"loss": 0.5977, "learning_rate": 0.00015502951988414837, "epoch": 0.97, "step": 26140}
+{"loss": 0.6845, "learning_rate": 0.00015497382198952878, "epoch": 0.97, "step": 26150}
+{"loss": 0.623, "learning_rate": 0.00015491812409490922, "epoch": 0.97, "step": 26160}
+{"loss": 0.6241, "learning_rate": 0.00015486242620028963, "epoch": 0.97, "step": 26170}
+{"loss": 0.7736, "learning_rate": 0.00015480672830567004, "epoch": 0.97, "step": 26180}
+{"loss": 0.6809, "learning_rate": 0.00015475103041105045, "epoch": 0.97, "step": 26190}
+{"loss": 0.6846, "learning_rate": 0.00015469533251643086, "epoch": 0.97, "step": 26200}
+{"loss": 0.562, "learning_rate": 0.00015463963462181127, "epoch": 0.97, "step": 26210}
+{"loss": 0.6656, "learning_rate": 0.0001545839367271917, "epoch": 0.97, "step": 26220}
+{"loss": 0.5681, "learning_rate": 0.00015452823883257212, "epoch": 0.97, "step": 26230}
+{"loss": 0.749, "learning_rate": 0.00015447254093795253, "epoch": 0.97, "step": 26240}
+{"loss": 0.72, "learning_rate": 0.00015441684304333294, "epoch": 0.97, "step": 26250}
+{"loss": 0.7233, "learning_rate": 0.00015436114514871336, "epoch": 0.97, "step": 26260}
+{"loss": 0.658, "learning_rate": 0.00015430544725409377, "epoch": 0.97, "step": 26270}
+{"loss": 0.6725, "learning_rate": 0.00015424974935947418, "epoch": 0.97, "step": 26280}
+{"loss": 0.6724, "learning_rate": 0.00015419405146485462, "epoch": 0.97, "step": 26290}
+{"loss": 0.6132, "learning_rate": 0.00015413835357023503, "epoch": 0.97, "step": 26300}
+{"loss": 0.7725, "learning_rate": 0.00015408265567561544, "epoch": 0.98, "step": 26310}
+{"loss": 0.6893, "learning_rate": 0.00015402695778099585, "epoch": 0.98, "step": 26320}
+{"loss": 0.5694, "learning_rate": 0.0001539712598863763, "epoch": 0.98, "step": 26330}
+{"loss": 0.6053, "learning_rate": 0.0001539155619917567, "epoch": 0.98, "step": 26340}
+{"loss": 0.79, "learning_rate": 0.00015385986409713714, "epoch": 0.98, "step": 26350}
+{"loss": 0.6865, "learning_rate": 0.00015380416620251755, "epoch": 0.98, "step": 26360}
+{"loss": 0.5849, "learning_rate": 0.00015374846830789796, "epoch": 0.98, "step": 26370}
+{"loss": 0.5446, "learning_rate": 0.00015369277041327837, "epoch": 0.98, "step": 26380}
+{"loss": 0.6265, "learning_rate": 0.00015363707251865878, "epoch": 0.98, "step": 26390}
+{"loss": 0.5822, "learning_rate": 0.0001535813746240392, "epoch": 0.98, "step": 26400}
+{"loss": 0.6995, "learning_rate": 0.0001535256767294196, "epoch": 0.98, "step": 26410}
+{"loss": 0.6874, "learning_rate": 0.00015346997883480004, "epoch": 0.98, "step": 26420}
+{"loss": 0.5584, "learning_rate": 0.00015341428094018045, "epoch": 0.98, "step": 26430}
+{"loss": 0.6684, "learning_rate": 0.00015335858304556086, "epoch": 0.98, "step": 26440}
+{"loss": 0.7603, "learning_rate": 0.00015330288515094128, "epoch": 0.98, "step": 26450}
+{"loss": 0.6375, "learning_rate": 0.0001532471872563217, "epoch": 0.98, "step": 26460}
+{"loss": 0.7927, "learning_rate": 0.0001531914893617021, "epoch": 0.98, "step": 26470}
+{"loss": 0.7341, "learning_rate": 0.00015313579146708254, "epoch": 0.98, "step": 26480}
+{"loss": 0.6145, "learning_rate": 0.00015308009357246295, "epoch": 0.98, "step": 26490}
+{"loss": 0.7039, "learning_rate": 0.00015302439567784336, "epoch": 0.98, "step": 26500}
+{"loss": 0.565, "learning_rate": 0.00015296869778322377, "epoch": 0.98, "step": 26510}
+{"loss": 0.7221, "learning_rate": 0.00015291299988860418, "epoch": 0.98, "step": 26520}
+{"loss": 0.5973, "learning_rate": 0.0001528573019939846, "epoch": 0.98, "step": 26530}
+{"loss": 0.6814, "learning_rate": 0.000152801604099365, "epoch": 0.98, "step": 26540}
+{"loss": 0.6685, "learning_rate": 0.00015274590620474547, "epoch": 0.98, "step": 26550}
+{"loss": 0.5877, "learning_rate": 0.00015269020831012588, "epoch": 0.98, "step": 26560}
+{"loss": 0.7141, "learning_rate": 0.0001526345104155063, "epoch": 0.98, "step": 26570}
+{"loss": 0.6058, "learning_rate": 0.0001525788125208867, "epoch": 0.99, "step": 26580}
+{"loss": 0.7058, "learning_rate": 0.0001525231146262671, "epoch": 0.99, "step": 26590}
+{"loss": 0.654, "learning_rate": 0.00015246741673164752, "epoch": 0.99, "step": 26600}
+{"loss": 0.615, "learning_rate": 0.00015241171883702796, "epoch": 0.99, "step": 26610}
+{"loss": 0.6208, "learning_rate": 0.00015235602094240837, "epoch": 0.99, "step": 26620}
+{"loss": 0.7558, "learning_rate": 0.00015230032304778878, "epoch": 0.99, "step": 26630}
+{"loss": 0.584, "learning_rate": 0.0001522446251531692, "epoch": 0.99, "step": 26640}
+{"loss": 0.571, "learning_rate": 0.0001521889272585496, "epoch": 0.99, "step": 26650}
+{"loss": 0.6147, "learning_rate": 0.00015213322936393002, "epoch": 0.99, "step": 26660}
+{"loss": 0.5617, "learning_rate": 0.00015207753146931043, "epoch": 0.99, "step": 26670}
+{"loss": 0.6879, "learning_rate": 0.00015202183357469087, "epoch": 0.99, "step": 26680}
+{"loss": 0.5883, "learning_rate": 0.00015196613568007128, "epoch": 0.99, "step": 26690}
+{"loss": 0.6871, "learning_rate": 0.0001519104377854517, "epoch": 0.99, "step": 26700}
+{"loss": 0.6884, "learning_rate": 0.0001518547398908321, "epoch": 0.99, "step": 26710}
+{"loss": 0.6958, "learning_rate": 0.0001517990419962125, "epoch": 0.99, "step": 26720}
+{"loss": 0.7593, "learning_rate": 0.00015174334410159292, "epoch": 0.99, "step": 26730}
+{"loss": 0.6588, "learning_rate": 0.0001516876462069734, "epoch": 0.99, "step": 26740}
+{"loss": 0.6906, "learning_rate": 0.0001516319483123538, "epoch": 0.99, "step": 26750}
+{"loss": 0.5798, "learning_rate": 0.0001515762504177342, "epoch": 0.99, "step": 26760}
+{"loss": 0.5316, "learning_rate": 0.00015152055252311462, "epoch": 0.99, "step": 26770}
+{"loss": 0.7071, "learning_rate": 0.00015146485462849503, "epoch": 0.99, "step": 26780}
+{"loss": 0.6039, "learning_rate": 0.00015140915673387544, "epoch": 0.99, "step": 26790}
+{"loss": 0.6719, "learning_rate": 0.00015135345883925586, "epoch": 0.99, "step": 26800}
+{"loss": 0.5624, "learning_rate": 0.0001512977609446363, "epoch": 0.99, "step": 26810}
+{"loss": 0.6667, "learning_rate": 0.0001512420630500167, "epoch": 0.99, "step": 26820}
+{"loss": 0.6293, "learning_rate": 0.00015118636515539712, "epoch": 0.99, "step": 26830}
+{"loss": 0.564, "learning_rate": 0.00015113066726077753, "epoch": 0.99, "step": 26840}
+{"loss": 0.6243, "learning_rate": 0.00015107496936615794, "epoch": 1.0, "step": 26850}
+{"loss": 0.707, "learning_rate": 0.00015101927147153835, "epoch": 1.0, "step": 26860}
+{"loss": 0.6507, "learning_rate": 0.0001509635735769188, "epoch": 1.0, "step": 26870}
+{"loss": 0.5755, "learning_rate": 0.0001509078756822992, "epoch": 1.0, "step": 26880}
+{"loss": 0.658, "learning_rate": 0.0001508521777876796, "epoch": 1.0, "step": 26890}
+{"loss": 0.7153, "learning_rate": 0.00015079647989306002, "epoch": 1.0, "step": 26900}
+{"loss": 0.6214, "learning_rate": 0.00015074078199844043, "epoch": 1.0, "step": 26910}
+{"loss": 0.6069, "learning_rate": 0.00015068508410382084, "epoch": 1.0, "step": 26920}
+{"loss": 0.5345, "learning_rate": 0.00015062938620920125, "epoch": 1.0, "step": 26930}
+{"loss": 0.7243, "learning_rate": 0.00015057368831458172, "epoch": 1.0, "step": 26940}
+{"loss": 0.7971, "learning_rate": 0.00015051799041996213, "epoch": 1.0, "step": 26950}
+{"loss": 0.5465, "learning_rate": 0.00015046229252534254, "epoch": 1.0, "step": 26960}
+{"loss": 0.5908, "learning_rate": 0.00015040659463072295, "epoch": 1.0, "step": 26970}
+{"loss": 0.668, "learning_rate": 0.00015035089673610336, "epoch": 1.0, "step": 26980}
+{"loss": 0.4634, "learning_rate": 0.00015029519884148378, "epoch": 1.0, "step": 26990}
+{"loss": 0.7426, "learning_rate": 0.00015023950094686421, "epoch": 1.0, "step": 27000}
+{"loss": 0.5737, "learning_rate": 0.00015018380305224462, "epoch": 1.0, "step": 27010}
+{"loss": 0.6468, "learning_rate": 0.00015012810515762504, "epoch": 1.0, "step": 27020}
+{"loss": 0.5475, "learning_rate": 0.00015007240726300545, "epoch": 1.0, "step": 27030}
+{"loss": 0.7392, "learning_rate": 0.00015001670936838586, "epoch": 1.0, "step": 27040}
+{"loss": 0.7629, "learning_rate": 0.0001499610114737663, "epoch": 1.0, "step": 27050}
+{"loss": 0.6523, "learning_rate": 0.0001499053135791467, "epoch": 1.0, "step": 27060}
+{"loss": 0.6695, "learning_rate": 0.00014984961568452712, "epoch": 1.0, "step": 27070}
+{"loss": 0.447, "learning_rate": 0.00014979391778990753, "epoch": 1.0, "step": 27080}
+{"loss": 0.6267, "learning_rate": 0.00014973821989528794, "epoch": 1.0, "step": 27090}
+{"loss": 0.6354, "learning_rate": 0.00014968252200066835, "epoch": 1.0, "step": 27100}
+{"loss": 0.5283, "learning_rate": 0.00014962682410604876, "epoch": 1.0, "step": 27110}
+{"loss": 0.551, "learning_rate": 0.0001495711262114292, "epoch": 1.01, "step": 27120}
+{"loss": 0.7446, "learning_rate": 0.0001495154283168096, "epoch": 1.01, "step": 27130}
+{"loss": 0.6156, "learning_rate": 0.00014945973042219002, "epoch": 1.01, "step": 27140}
+{"loss": 0.6953, "learning_rate": 0.00014940403252757046, "epoch": 1.01, "step": 27150}
+{"loss": 0.5751, "learning_rate": 0.00014934833463295087, "epoch": 1.01, "step": 27160}
+{"loss": 0.52, "learning_rate": 0.00014929263673833128, "epoch": 1.01, "step": 27170}
+{"loss": 0.5435, "learning_rate": 0.0001492369388437117, "epoch": 1.01, "step": 27180}
+{"loss": 0.6627, "learning_rate": 0.0001491812409490921, "epoch": 1.01, "step": 27190}
+{"loss": 0.6555, "learning_rate": 0.00014912554305447252, "epoch": 1.01, "step": 27200}
+{"loss": 0.6808, "learning_rate": 0.00014906984515985296, "epoch": 1.01, "step": 27210}
+{"loss": 0.5402, "learning_rate": 0.00014901414726523337, "epoch": 1.01, "step": 27220}
+{"loss": 0.6724, "learning_rate": 0.00014895844937061378, "epoch": 1.01, "step": 27230}
+{"loss": 0.6722, "learning_rate": 0.0001489027514759942, "epoch": 1.01, "step": 27240}
+{"loss": 0.5147, "learning_rate": 0.00014884705358137463, "epoch": 1.01, "step": 27250}
+{"loss": 0.8395, "learning_rate": 0.00014879135568675504, "epoch": 1.01, "step": 27260}
+{"loss": 0.605, "learning_rate": 0.00014873565779213545, "epoch": 1.01, "step": 27270}
+{"loss": 0.6245, "learning_rate": 0.00014867995989751586, "epoch": 1.01, "step": 27280}
+{"loss": 0.5898, "learning_rate": 0.00014862426200289627, "epoch": 1.01, "step": 27290}
+{"loss": 0.557, "learning_rate": 0.00014856856410827668, "epoch": 1.01, "step": 27300}
+{"loss": 0.704, "learning_rate": 0.00014851286621365712, "epoch": 1.01, "step": 27310}
+{"loss": 0.4358, "learning_rate": 0.00014845716831903753, "epoch": 1.01, "step": 27320}
+{"loss": 0.5128, "learning_rate": 0.00014840147042441794, "epoch": 1.01, "step": 27330}
+{"loss": 0.5276, "learning_rate": 0.00014834577252979835, "epoch": 1.01, "step": 27340}
+{"loss": 0.6491, "learning_rate": 0.00014829007463517877, "epoch": 1.01, "step": 27350}
+{"loss": 0.6062, "learning_rate": 0.0001482343767405592, "epoch": 1.01, "step": 27360}
+{"loss": 0.759, "learning_rate": 0.00014817867884593962, "epoch": 1.01, "step": 27370}
+{"loss": 0.5563, "learning_rate": 0.00014812298095132003, "epoch": 1.01, "step": 27380}
+{"loss": 0.764, "learning_rate": 0.00014806728305670044, "epoch": 1.02, "step": 27390}
+{"loss": 0.4679, "learning_rate": 0.00014801158516208085, "epoch": 1.02, "step": 27400}
+{"loss": 0.5921, "learning_rate": 0.0001479558872674613, "epoch": 1.02, "step": 27410}
+{"loss": 0.6356, "learning_rate": 0.0001479001893728417, "epoch": 1.02, "step": 27420}
+{"loss": 0.5951, "learning_rate": 0.0001478444914782221, "epoch": 1.02, "step": 27430}
+{"loss": 0.5872, "learning_rate": 0.00014778879358360252, "epoch": 1.02, "step": 27440}
+{"loss": 0.5969, "learning_rate": 0.00014773309568898293, "epoch": 1.02, "step": 27450}
+{"loss": 0.5302, "learning_rate": 0.00014767739779436334, "epoch": 1.02, "step": 27460}
+{"loss": 0.6272, "learning_rate": 0.00014762169989974378, "epoch": 1.02, "step": 27470}
+{"loss": 0.563, "learning_rate": 0.0001475660020051242, "epoch": 1.02, "step": 27480}
+{"loss": 0.6368, "learning_rate": 0.0001475103041105046, "epoch": 1.02, "step": 27490}
+{"loss": 0.5762, "learning_rate": 0.00014745460621588501, "epoch": 1.02, "step": 27500}
+{"loss": 0.6564, "learning_rate": 0.00014739890832126545, "epoch": 1.02, "step": 27510}
+{"loss": 0.747, "learning_rate": 0.00014734321042664586, "epoch": 1.02, "step": 27520}
+{"loss": 0.4904, "learning_rate": 0.00014728751253202627, "epoch": 1.02, "step": 27530}
+{"loss": 0.5357, "learning_rate": 0.00014723181463740669, "epoch": 1.02, "step": 27540}
+{"loss": 0.6118, "learning_rate": 0.0001471761167427871, "epoch": 1.02, "step": 27550}
+{"loss": 0.6067, "learning_rate": 0.0001471204188481675, "epoch": 1.02, "step": 27560}
+{"loss": 0.5896, "learning_rate": 0.00014706472095354795, "epoch": 1.02, "step": 27570}
+{"loss": 0.5464, "learning_rate": 0.00014700902305892836, "epoch": 1.02, "step": 27580}
+{"loss": 0.6674, "learning_rate": 0.00014695332516430877, "epoch": 1.02, "step": 27590}
+{"loss": 0.6726, "learning_rate": 0.0001468976272696892, "epoch": 1.02, "step": 27600}
+{"loss": 0.5646, "learning_rate": 0.00014684192937506962, "epoch": 1.02, "step": 27610}
+{"loss": 0.5054, "learning_rate": 0.00014678623148045003, "epoch": 1.02, "step": 27620}
+{"loss": 0.53, "learning_rate": 0.00014673053358583044, "epoch": 1.02, "step": 27630}
+{"loss": 0.6189, "learning_rate": 0.00014667483569121085, "epoch": 1.02, "step": 27640}
+{"loss": 0.6531, "learning_rate": 0.00014661913779659126, "epoch": 1.02, "step": 27650}
+{"loss": 0.6671, "learning_rate": 0.00014656343990197167, "epoch": 1.03, "step": 27660}
+{"loss": 0.5816, "learning_rate": 0.0001465077420073521, "epoch": 1.03, "step": 27670}
+{"loss": 0.6263, "learning_rate": 0.00014645204411273252, "epoch": 1.03, "step": 27680}
+{"loss": 0.6456, "learning_rate": 0.00014639634621811293, "epoch": 1.03, "step": 27690}
+{"loss": 0.5815, "learning_rate": 0.00014634064832349337, "epoch": 1.03, "step": 27700}
+{"loss": 0.6021, "learning_rate": 0.00014628495042887378, "epoch": 1.03, "step": 27710}
+{"loss": 0.6195, "learning_rate": 0.0001462292525342542, "epoch": 1.03, "step": 27720}
+{"loss": 0.576, "learning_rate": 0.0001461735546396346, "epoch": 1.03, "step": 27730}
+{"loss": 0.6761, "learning_rate": 0.00014611785674501502, "epoch": 1.03, "step": 27740}
+{"loss": 0.4831, "learning_rate": 0.00014606215885039543, "epoch": 1.03, "step": 27750}
+{"loss": 0.5613, "learning_rate": 0.00014600646095577587, "epoch": 1.03, "step": 27760}
+{"loss": 0.5572, "learning_rate": 0.00014595076306115628, "epoch": 1.03, "step": 27770}
+{"loss": 0.5268, "learning_rate": 0.0001458950651665367, "epoch": 1.03, "step": 27780}
+{"loss": 0.6035, "learning_rate": 0.00014584493706137908, "epoch": 1.03, "step": 27790}
+{"loss": 0.5981, "learning_rate": 0.00014578923916675949, "epoch": 1.03, "step": 27800}
+{"loss": 0.6424, "learning_rate": 0.0001457335412721399, "epoch": 1.03, "step": 27810}
+{"loss": 0.4824, "learning_rate": 0.0001456778433775203, "epoch": 1.03, "step": 27820}
+{"loss": 0.6069, "learning_rate": 0.00014562214548290072, "epoch": 1.03, "step": 27830}
+{"loss": 0.497, "learning_rate": 0.00014556644758828116, "epoch": 1.03, "step": 27840}
+{"loss": 0.5539, "learning_rate": 0.00014551074969366157, "epoch": 1.03, "step": 27850}
+{"loss": 0.6126, "learning_rate": 0.00014545505179904198, "epoch": 1.03, "step": 27860}
+{"loss": 0.5176, "learning_rate": 0.00014539935390442242, "epoch": 1.03, "step": 27870}
+{"loss": 0.6168, "learning_rate": 0.00014534365600980283, "epoch": 1.03, "step": 27880}
+{"loss": 0.5872, "learning_rate": 0.00014528795811518324, "epoch": 1.03, "step": 27890}
+{"loss": 0.6517, "learning_rate": 0.00014523226022056365, "epoch": 1.03, "step": 27900}
+{"loss": 0.5029, "learning_rate": 0.00014517656232594406, "epoch": 1.03, "step": 27910}
+{"loss": 0.5582, "learning_rate": 0.00014512086443132447, "epoch": 1.03, "step": 27920}
+{"loss": 0.515, "learning_rate": 0.0001450651665367049, "epoch": 1.04, "step": 27930}
+{"loss": 0.5168, "learning_rate": 0.00014500946864208532, "epoch": 1.04, "step": 27940}
+{"loss": 0.5861, "learning_rate": 0.00014495377074746573, "epoch": 1.04, "step": 27950}
+{"loss": 0.6858, "learning_rate": 0.00014489807285284615, "epoch": 1.04, "step": 27960}
+{"loss": 0.5829, "learning_rate": 0.00014484237495822656, "epoch": 1.04, "step": 27970}
+{"loss": 0.5791, "learning_rate": 0.000144786677063607, "epoch": 1.04, "step": 27980}
+{"loss": 0.6624, "learning_rate": 0.0001447309791689874, "epoch": 1.04, "step": 27990}
+{"loss": 0.6267, "learning_rate": 0.00014467528127436782, "epoch": 1.04, "step": 28000}
+{"loss": 0.5421, "learning_rate": 0.00014461958337974823, "epoch": 1.04, "step": 28010}
+{"loss": 0.5864, "learning_rate": 0.00014456388548512864, "epoch": 1.04, "step": 28020}
+{"loss": 0.5509, "learning_rate": 0.00014450818759050908, "epoch": 1.04, "step": 28030}
+{"loss": 0.6254, "learning_rate": 0.0001444524896958895, "epoch": 1.04, "step": 28040}
+{"loss": 0.6194, "learning_rate": 0.0001443967918012699, "epoch": 1.04, "step": 28050}
+{"loss": 0.5668, "learning_rate": 0.0001443410939066503, "epoch": 1.04, "step": 28060}
+{"loss": 0.6672, "learning_rate": 0.00014428539601203072, "epoch": 1.04, "step": 28070}
+{"loss": 0.5642, "learning_rate": 0.00014422969811741113, "epoch": 1.04, "step": 28080}
+{"loss": 0.639, "learning_rate": 0.00014417400022279154, "epoch": 1.04, "step": 28090}
+{"loss": 0.5227, "learning_rate": 0.00014411830232817198, "epoch": 1.04, "step": 28100}
+{"loss": 0.6036, "learning_rate": 0.0001440626044335524, "epoch": 1.04, "step": 28110}
+{"loss": 0.5755, "learning_rate": 0.0001440069065389328, "epoch": 1.04, "step": 28120}
+{"loss": 0.6136, "learning_rate": 0.00014395120864431324, "epoch": 1.04, "step": 28130}
+{"loss": 0.6739, "learning_rate": 0.00014389551074969365, "epoch": 1.04, "step": 28140}
+{"loss": 0.7447, "learning_rate": 0.00014383981285507407, "epoch": 1.04, "step": 28150}
+{"loss": 0.5471, "learning_rate": 0.00014378411496045448, "epoch": 1.04, "step": 28160}
+{"loss": 0.5419, "learning_rate": 0.0001437284170658349, "epoch": 1.04, "step": 28170}
+{"loss": 0.59, "learning_rate": 0.0001436727191712153, "epoch": 1.04, "step": 28180}
+{"loss": 0.6926, "learning_rate": 0.00014361702127659574, "epoch": 1.04, "step": 28190}
+{"loss": 0.5994, "learning_rate": 0.00014356132338197615, "epoch": 1.05, "step": 28200}
+{"loss": 0.5359, "learning_rate": 0.00014350562548735656, "epoch": 1.05, "step": 28210}
+{"loss": 0.6353, "learning_rate": 0.00014344992759273697, "epoch": 1.05, "step": 28220}
+{"loss": 0.5661, "learning_rate": 0.0001433942296981174, "epoch": 1.05, "step": 28230}
+{"loss": 0.5923, "learning_rate": 0.00014333853180349782, "epoch": 1.05, "step": 28240}
+{"loss": 0.6818, "learning_rate": 0.00014328283390887823, "epoch": 1.05, "step": 28250}
+{"loss": 0.6298, "learning_rate": 0.00014322713601425864, "epoch": 1.05, "step": 28260}
+{"loss": 0.653, "learning_rate": 0.00014317143811963905, "epoch": 1.05, "step": 28270}
+{"loss": 0.5953, "learning_rate": 0.00014311574022501946, "epoch": 1.05, "step": 28280}
+{"loss": 0.5873, "learning_rate": 0.0001430600423303999, "epoch": 1.05, "step": 28290}
+{"loss": 0.5688, "learning_rate": 0.00014300434443578031, "epoch": 1.05, "step": 28300}
+{"loss": 0.6153, "learning_rate": 0.00014294864654116073, "epoch": 1.05, "step": 28310}
+{"loss": 0.622, "learning_rate": 0.00014289294864654116, "epoch": 1.05, "step": 28320}
+{"loss": 0.4433, "learning_rate": 0.00014283725075192157, "epoch": 1.05, "step": 28330}
+{"loss": 0.6258, "learning_rate": 0.00014278155285730199, "epoch": 1.05, "step": 28340}
+{"loss": 0.6222, "learning_rate": 0.0001427258549626824, "epoch": 1.05, "step": 28350}
+{"loss": 0.7373, "learning_rate": 0.0001426701570680628, "epoch": 1.05, "step": 28360}
+{"loss": 0.6533, "learning_rate": 0.00014261445917344322, "epoch": 1.05, "step": 28370}
+{"loss": 0.6338, "learning_rate": 0.00014255876127882363, "epoch": 1.05, "step": 28380}
+{"loss": 0.6207, "learning_rate": 0.00014250306338420407, "epoch": 1.05, "step": 28390}
+{"loss": 0.5709, "learning_rate": 0.00014244736548958448, "epoch": 1.05, "step": 28400}
+{"loss": 0.5064, "learning_rate": 0.0001423916675949649, "epoch": 1.05, "step": 28410}
+{"loss": 0.5504, "learning_rate": 0.00014233596970034533, "epoch": 1.05, "step": 28420}
+{"loss": 0.5052, "learning_rate": 0.00014228027180572574, "epoch": 1.05, "step": 28430}
+{"loss": 0.5614, "learning_rate": 0.00014222457391110615, "epoch": 1.05, "step": 28440}
+{"loss": 0.751, "learning_rate": 0.00014216887601648656, "epoch": 1.05, "step": 28450}
+{"loss": 0.5501, "learning_rate": 0.00014211317812186697, "epoch": 1.05, "step": 28460}
+{"loss": 0.5157, "learning_rate": 0.00014205748022724738, "epoch": 1.06, "step": 28470}
+{"loss": 0.6579, "learning_rate": 0.0001420017823326278, "epoch": 1.06, "step": 28480}
+{"loss": 0.5991, "learning_rate": 0.00014194608443800823, "epoch": 1.06, "step": 28490}
+{"loss": 0.621, "learning_rate": 0.00014189038654338865, "epoch": 1.06, "step": 28500}
+{"loss": 0.6897, "learning_rate": 0.00014183468864876906, "epoch": 1.06, "step": 28510}
+{"loss": 0.6427, "learning_rate": 0.0001417789907541495, "epoch": 1.06, "step": 28520}
+{"loss": 0.6338, "learning_rate": 0.0001417232928595299, "epoch": 1.06, "step": 28530}
+{"loss": 0.6423, "learning_rate": 0.00014166759496491032, "epoch": 1.06, "step": 28540}
+{"loss": 0.6832, "learning_rate": 0.00014161189707029073, "epoch": 1.06, "step": 28550}
+{"loss": 0.6847, "learning_rate": 0.00014155619917567114, "epoch": 1.06, "step": 28560}
+{"loss": 0.6373, "learning_rate": 0.00014150050128105155, "epoch": 1.06, "step": 28570}
+{"loss": 0.6729, "learning_rate": 0.000141444803386432, "epoch": 1.06, "step": 28580}
+{"loss": 0.6302, "learning_rate": 0.0001413891054918124, "epoch": 1.06, "step": 28590}
+{"loss": 0.5747, "learning_rate": 0.0001413334075971928, "epoch": 1.06, "step": 28600}
+{"loss": 0.5689, "learning_rate": 0.00014127770970257325, "epoch": 1.06, "step": 28610}
+{"loss": 0.7226, "learning_rate": 0.00014122201180795366, "epoch": 1.06, "step": 28620}
+{"loss": 0.6597, "learning_rate": 0.00014116631391333407, "epoch": 1.06, "step": 28630}
+{"loss": 0.6318, "learning_rate": 0.00014111061601871448, "epoch": 1.06, "step": 28640}
+{"loss": 0.7549, "learning_rate": 0.0001410549181240949, "epoch": 1.06, "step": 28650}
+{"loss": 0.6374, "learning_rate": 0.0001409992202294753, "epoch": 1.06, "step": 28660}
+{"loss": 0.6357, "learning_rate": 0.00014094352233485572, "epoch": 1.06, "step": 28670}
+{"loss": 0.6386, "learning_rate": 0.00014088782444023615, "epoch": 1.06, "step": 28680}
+{"loss": 0.7495, "learning_rate": 0.00014083212654561657, "epoch": 1.06, "step": 28690}
+{"loss": 0.5263, "learning_rate": 0.00014077642865099698, "epoch": 1.06, "step": 28700}
+{"loss": 0.5395, "learning_rate": 0.00014072073075637741, "epoch": 1.06, "step": 28710}
+{"loss": 0.6153, "learning_rate": 0.00014066503286175783, "epoch": 1.06, "step": 28720}
+{"loss": 0.5783, "learning_rate": 0.00014060933496713824, "epoch": 1.06, "step": 28730}
+{"loss": 0.5446, "learning_rate": 0.00014055363707251865, "epoch": 1.07, "step": 28740}
+{"loss": 0.6922, "learning_rate": 0.00014049793917789906, "epoch": 1.07, "step": 28750}
+{"loss": 0.4609, "learning_rate": 0.00014044224128327947, "epoch": 1.07, "step": 28760}
+{"loss": 0.6839, "learning_rate": 0.00014038654338865988, "epoch": 1.07, "step": 28770}
+{"loss": 0.6182, "learning_rate": 0.00014033084549404032, "epoch": 1.07, "step": 28780}
+{"loss": 0.7257, "learning_rate": 0.00014027514759942073, "epoch": 1.07, "step": 28790}
+{"loss": 0.6834, "learning_rate": 0.00014021944970480114, "epoch": 1.07, "step": 28800}
+{"loss": 0.5532, "learning_rate": 0.00014016375181018158, "epoch": 1.07, "step": 28810}
+{"loss": 0.6574, "learning_rate": 0.000140108053915562, "epoch": 1.07, "step": 28820}
+{"loss": 0.58, "learning_rate": 0.0001400523560209424, "epoch": 1.07, "step": 28830}
+{"loss": 0.5084, "learning_rate": 0.0001399966581263228, "epoch": 1.07, "step": 28840}
+{"loss": 0.6035, "learning_rate": 0.00013994096023170322, "epoch": 1.07, "step": 28850}
+{"loss": 0.5978, "learning_rate": 0.00013988526233708364, "epoch": 1.07, "step": 28860}
+{"loss": 0.6123, "learning_rate": 0.00013982956444246407, "epoch": 1.07, "step": 28870}
+{"loss": 0.5973, "learning_rate": 0.00013977386654784449, "epoch": 1.07, "step": 28880}
+{"loss": 0.545, "learning_rate": 0.0001397181686532249, "epoch": 1.07, "step": 28890}
+{"loss": 0.6065, "learning_rate": 0.0001396624707586053, "epoch": 1.07, "step": 28900}
+{"loss": 0.5697, "learning_rate": 0.00013960677286398575, "epoch": 1.07, "step": 28910}
+{"loss": 0.6563, "learning_rate": 0.00013955107496936616, "epoch": 1.07, "step": 28920}
+{"loss": 0.5535, "learning_rate": 0.00013949537707474657, "epoch": 1.07, "step": 28930}
+{"loss": 0.5575, "learning_rate": 0.00013943967918012698, "epoch": 1.07, "step": 28940}
+{"loss": 0.5724, "learning_rate": 0.0001393839812855074, "epoch": 1.07, "step": 28950}
+{"loss": 0.5215, "learning_rate": 0.0001393282833908878, "epoch": 1.07, "step": 28960}
+{"loss": 0.6067, "learning_rate": 0.0001392781552857302, "epoch": 1.07, "step": 28970}
+{"loss": 0.6617, "learning_rate": 0.0001392224573911106, "epoch": 1.07, "step": 28980}
+{"loss": 0.6187, "learning_rate": 0.000139166759496491, "epoch": 1.07, "step": 28990}
+{"loss": 0.5893, "learning_rate": 0.00013911106160187142, "epoch": 1.07, "step": 29000}
+{"loss": 0.6454, "learning_rate": 0.00013905536370725186, "epoch": 1.08, "step": 29010}
+{"loss": 0.5571, "learning_rate": 0.00013899966581263227, "epoch": 1.08, "step": 29020}
+{"loss": 0.7921, "learning_rate": 0.00013894396791801268, "epoch": 1.08, "step": 29030}
+{"loss": 0.5683, "learning_rate": 0.00013888827002339312, "epoch": 1.08, "step": 29040}
+{"loss": 0.5369, "learning_rate": 0.00013883257212877353, "epoch": 1.08, "step": 29050}
+{"loss": 0.5291, "learning_rate": 0.00013877687423415394, "epoch": 1.08, "step": 29060}
+{"loss": 0.6136, "learning_rate": 0.00013872117633953435, "epoch": 1.08, "step": 29070}
+{"loss": 0.569, "learning_rate": 0.00013866547844491476, "epoch": 1.08, "step": 29080}
+{"loss": 0.4386, "learning_rate": 0.00013860978055029518, "epoch": 1.08, "step": 29090}
+{"loss": 0.5527, "learning_rate": 0.0001385540826556756, "epoch": 1.08, "step": 29100}
+{"loss": 0.5459, "learning_rate": 0.00013849838476105602, "epoch": 1.08, "step": 29110}
+{"loss": 0.6416, "learning_rate": 0.00013844268686643644, "epoch": 1.08, "step": 29120}
+{"loss": 0.586, "learning_rate": 0.00013838698897181685, "epoch": 1.08, "step": 29130}
+{"loss": 0.4898, "learning_rate": 0.00013833129107719729, "epoch": 1.08, "step": 29140}
+{"loss": 0.6606, "learning_rate": 0.0001382755931825777, "epoch": 1.08, "step": 29150}
+{"loss": 0.7579, "learning_rate": 0.0001382198952879581, "epoch": 1.08, "step": 29160}
+{"loss": 0.7036, "learning_rate": 0.00013816419739333852, "epoch": 1.08, "step": 29170}
+{"loss": 0.5601, "learning_rate": 0.00013810849949871893, "epoch": 1.08, "step": 29180}
+{"loss": 0.4931, "learning_rate": 0.00013805280160409934, "epoch": 1.08, "step": 29190}
+{"loss": 0.6479, "learning_rate": 0.00013799710370947975, "epoch": 1.08, "step": 29200}
+{"loss": 0.61, "learning_rate": 0.0001379414058148602, "epoch": 1.08, "step": 29210}
+{"loss": 0.6225, "learning_rate": 0.0001378857079202406, "epoch": 1.08, "step": 29220}
+{"loss": 0.5489, "learning_rate": 0.000137830010025621, "epoch": 1.08, "step": 29230}
+{"loss": 0.5307, "learning_rate": 0.00013777431213100145, "epoch": 1.08, "step": 29240}
+{"loss": 0.6051, "learning_rate": 0.00013771861423638186, "epoch": 1.08, "step": 29250}
+{"loss": 0.604, "learning_rate": 0.00013766291634176227, "epoch": 1.08, "step": 29260}
+{"loss": 0.5571, "learning_rate": 0.00013760721844714268, "epoch": 1.08, "step": 29270}
+{"loss": 0.4529, "learning_rate": 0.0001375515205525231, "epoch": 1.09, "step": 29280}
+{"loss": 0.4945, "learning_rate": 0.0001374958226579035, "epoch": 1.09, "step": 29290}
+{"loss": 0.7455, "learning_rate": 0.00013744012476328394, "epoch": 1.09, "step": 29300}
+{"loss": 0.7226, "learning_rate": 0.00013738442686866436, "epoch": 1.09, "step": 29310}
+{"loss": 0.6678, "learning_rate": 0.00013732872897404477, "epoch": 1.09, "step": 29320}
+{"loss": 0.6338, "learning_rate": 0.00013727303107942518, "epoch": 1.09, "step": 29330}
+{"loss": 0.5416, "learning_rate": 0.00013721733318480562, "epoch": 1.09, "step": 29340}
+{"loss": 0.5477, "learning_rate": 0.00013716163529018603, "epoch": 1.09, "step": 29350}
+{"loss": 0.6874, "learning_rate": 0.00013710593739556644, "epoch": 1.09, "step": 29360}
+{"loss": 0.5968, "learning_rate": 0.00013705023950094685, "epoch": 1.09, "step": 29370}
+{"loss": 0.4746, "learning_rate": 0.00013699454160632726, "epoch": 1.09, "step": 29380}
+{"loss": 0.5656, "learning_rate": 0.00013693884371170767, "epoch": 1.09, "step": 29390}
+{"loss": 0.6235, "learning_rate": 0.0001368831458170881, "epoch": 1.09, "step": 29400}
+{"loss": 0.5645, "learning_rate": 0.00013682744792246852, "epoch": 1.09, "step": 29410}
+{"loss": 0.5845, "learning_rate": 0.00013677175002784893, "epoch": 1.09, "step": 29420}
+{"loss": 0.5449, "learning_rate": 0.00013671605213322937, "epoch": 1.09, "step": 29430}
+{"loss": 0.6532, "learning_rate": 0.00013666035423860978, "epoch": 1.09, "step": 29440}
+{"loss": 0.6428, "learning_rate": 0.0001366046563439902, "epoch": 1.09, "step": 29450}
+{"loss": 0.5571, "learning_rate": 0.0001365489584493706, "epoch": 1.09, "step": 29460}
+{"loss": 0.624, "learning_rate": 0.00013649326055475102, "epoch": 1.09, "step": 29470}
+{"loss": 0.5874, "learning_rate": 0.00013643756266013143, "epoch": 1.09, "step": 29480}
+{"loss": 0.5713, "learning_rate": 0.00013638186476551184, "epoch": 1.09, "step": 29490}
+{"loss": 0.6077, "learning_rate": 0.00013632616687089228, "epoch": 1.09, "step": 29500}
+{"loss": 0.5108, "learning_rate": 0.0001362704689762727, "epoch": 1.09, "step": 29510}
+{"loss": 0.6291, "learning_rate": 0.0001362147710816531, "epoch": 1.09, "step": 29520}
+{"loss": 0.6297, "learning_rate": 0.00013615907318703354, "epoch": 1.09, "step": 29530}
+{"loss": 0.5852, "learning_rate": 0.00013610337529241395, "epoch": 1.09, "step": 29540}
+{"loss": 0.5487, "learning_rate": 0.00013604767739779436, "epoch": 1.1, "step": 29550}
+{"loss": 0.64, "learning_rate": 0.00013599197950317477, "epoch": 1.1, "step": 29560}
+{"loss": 0.559, "learning_rate": 0.00013593628160855518, "epoch": 1.1, "step": 29570}
+{"loss": 0.7788, "learning_rate": 0.0001358805837139356, "epoch": 1.1, "step": 29580}
+{"loss": 0.6607, "learning_rate": 0.00013582488581931603, "epoch": 1.1, "step": 29590}
+{"loss": 0.6098, "learning_rate": 0.00013576918792469644, "epoch": 1.1, "step": 29600}
+{"loss": 0.6405, "learning_rate": 0.00013571349003007685, "epoch": 1.1, "step": 29610}
+{"loss": 0.5299, "learning_rate": 0.00013565779213545726, "epoch": 1.1, "step": 29620}
+{"loss": 0.5513, "learning_rate": 0.0001356020942408377, "epoch": 1.1, "step": 29630}
+{"loss": 0.7151, "learning_rate": 0.0001355463963462181, "epoch": 1.1, "step": 29640}
+{"loss": 0.6107, "learning_rate": 0.00013549069845159852, "epoch": 1.1, "step": 29650}
+{"loss": 0.6224, "learning_rate": 0.00013543500055697894, "epoch": 1.1, "step": 29660}
+{"loss": 0.4986, "learning_rate": 0.00013537930266235935, "epoch": 1.1, "step": 29670}
+{"loss": 0.5785, "learning_rate": 0.00013532360476773976, "epoch": 1.1, "step": 29680}
+{"loss": 0.7123, "learning_rate": 0.0001352679068731202, "epoch": 1.1, "step": 29690}
+{"loss": 0.6334, "learning_rate": 0.0001352122089785006, "epoch": 1.1, "step": 29700}
+{"loss": 0.5445, "learning_rate": 0.00013515651108388102, "epoch": 1.1, "step": 29710}
+{"loss": 0.5722, "learning_rate": 0.00013510081318926143, "epoch": 1.1, "step": 29720}
+{"loss": 0.6186, "learning_rate": 0.00013504511529464184, "epoch": 1.1, "step": 29730}
+{"loss": 0.6591, "learning_rate": 0.00013498941740002225, "epoch": 1.1, "step": 29740}
+{"loss": 0.6334, "learning_rate": 0.0001349337195054027, "epoch": 1.1, "step": 29750}
+{"loss": 0.6241, "learning_rate": 0.0001348780216107831, "epoch": 1.1, "step": 29760}
+{"loss": 0.6335, "learning_rate": 0.0001348223237161635, "epoch": 1.1, "step": 29770}
+{"loss": 0.576, "learning_rate": 0.00013476662582154392, "epoch": 1.1, "step": 29780}
+{"loss": 0.6302, "learning_rate": 0.00013471092792692436, "epoch": 1.1, "step": 29790}
+{"loss": 0.5341, "learning_rate": 0.00013465523003230477, "epoch": 1.1, "step": 29800}
+{"loss": 0.5089, "learning_rate": 0.00013459953213768518, "epoch": 1.1, "step": 29810}
+{"loss": 0.5115, "learning_rate": 0.0001345438342430656, "epoch": 1.11, "step": 29820}
+{"loss": 0.5861, "learning_rate": 0.000134488136348446, "epoch": 1.11, "step": 29830}
+{"loss": 0.5592, "learning_rate": 0.00013443243845382642, "epoch": 1.11, "step": 29840}
+{"loss": 0.6329, "learning_rate": 0.00013437674055920686, "epoch": 1.11, "step": 29850}
+{"loss": 0.6253, "learning_rate": 0.00013432104266458727, "epoch": 1.11, "step": 29860}
+{"loss": 0.6283, "learning_rate": 0.00013426534476996768, "epoch": 1.11, "step": 29870}
+{"loss": 0.548, "learning_rate": 0.0001342096468753481, "epoch": 1.11, "step": 29880}
+{"loss": 0.6634, "learning_rate": 0.00013415394898072853, "epoch": 1.11, "step": 29890}
+{"loss": 0.6905, "learning_rate": 0.00013409825108610894, "epoch": 1.11, "step": 29900}
+{"loss": 0.6338, "learning_rate": 0.00013404255319148935, "epoch": 1.11, "step": 29910}
+{"loss": 0.6552, "learning_rate": 0.00013398685529686976, "epoch": 1.11, "step": 29920}
+{"loss": 0.5042, "learning_rate": 0.00013393115740225017, "epoch": 1.11, "step": 29930}
+{"loss": 0.5048, "learning_rate": 0.00013387545950763058, "epoch": 1.11, "step": 29940}
+{"loss": 0.5588, "learning_rate": 0.00013381976161301102, "epoch": 1.11, "step": 29950}
+{"loss": 0.6686, "learning_rate": 0.00013376406371839143, "epoch": 1.11, "step": 29960}
+{"loss": 0.6719, "learning_rate": 0.00013370836582377184, "epoch": 1.11, "step": 29970}
+{"loss": 0.4942, "learning_rate": 0.00013365266792915228, "epoch": 1.11, "step": 29980}
+{"loss": 0.5265, "learning_rate": 0.0001335969700345327, "epoch": 1.11, "step": 29990}
+{"loss": 0.499, "learning_rate": 0.0001335412721399131, "epoch": 1.11, "step": 30000}
+{"loss": 0.6948, "learning_rate": 0.00013348557424529351, "epoch": 1.11, "step": 30010}
+{"loss": 0.6221, "learning_rate": 0.00013342987635067393, "epoch": 1.11, "step": 30020}
+{"loss": 0.6036, "learning_rate": 0.00013337417845605434, "epoch": 1.11, "step": 30030}
+{"loss": 0.6215, "learning_rate": 0.00013331848056143475, "epoch": 1.11, "step": 30040}
+{"loss": 0.6551, "learning_rate": 0.0001332627826668152, "epoch": 1.11, "step": 30050}
+{"loss": 0.6517, "learning_rate": 0.0001332070847721956, "epoch": 1.11, "step": 30060}
+{"loss": 0.6602, "learning_rate": 0.000133151386877576, "epoch": 1.11, "step": 30070}
+{"loss": 0.589, "learning_rate": 0.00013309568898295645, "epoch": 1.11, "step": 30080}
+{"loss": 0.4997, "learning_rate": 0.00013303999108833686, "epoch": 1.12, "step": 30090}
+{"loss": 0.6696, "learning_rate": 0.00013298429319371727, "epoch": 1.12, "step": 30100}
+{"loss": 0.6318, "learning_rate": 0.00013292859529909768, "epoch": 1.12, "step": 30110}
+{"loss": 0.5285, "learning_rate": 0.0001328728974044781, "epoch": 1.12, "step": 30120}
+{"loss": 0.6166, "learning_rate": 0.0001328171995098585, "epoch": 1.12, "step": 30130}
+{"loss": 0.6143, "learning_rate": 0.00013276150161523891, "epoch": 1.12, "step": 30140}
+{"loss": 0.7933, "learning_rate": 0.00013270580372061935, "epoch": 1.12, "step": 30150}
+{"loss": 0.5831, "learning_rate": 0.00013265010582599976, "epoch": 1.12, "step": 30160}
+{"loss": 0.5665, "learning_rate": 0.00013259440793138017, "epoch": 1.12, "step": 30170}
+{"loss": 0.5789, "learning_rate": 0.0001325387100367606, "epoch": 1.12, "step": 30180}
+{"loss": 0.6259, "learning_rate": 0.00013248301214214102, "epoch": 1.12, "step": 30190}
+{"loss": 0.5455, "learning_rate": 0.00013242731424752143, "epoch": 1.12, "step": 30200}
+{"loss": 0.5899, "learning_rate": 0.00013237161635290185, "epoch": 1.12, "step": 30210}
+{"loss": 0.532, "learning_rate": 0.00013231591845828226, "epoch": 1.12, "step": 30220}
+{"loss": 0.5555, "learning_rate": 0.00013226022056366267, "epoch": 1.12, "step": 30230}
+{"loss": 0.5188, "learning_rate": 0.0001322045226690431, "epoch": 1.12, "step": 30240}
+{"loss": 0.6565, "learning_rate": 0.00013214882477442352, "epoch": 1.12, "step": 30250}
+{"loss": 0.6032, "learning_rate": 0.00013209312687980393, "epoch": 1.12, "step": 30260}
+{"loss": 0.5055, "learning_rate": 0.00013203742898518434, "epoch": 1.12, "step": 30270}
+{"loss": 0.6185, "learning_rate": 0.00013198173109056478, "epoch": 1.12, "step": 30280}
+{"loss": 0.5907, "learning_rate": 0.0001319260331959452, "epoch": 1.12, "step": 30290}
+{"loss": 0.5669, "learning_rate": 0.0001318703353013256, "epoch": 1.12, "step": 30300}
+{"loss": 0.6931, "learning_rate": 0.000131814637406706, "epoch": 1.12, "step": 30310}
+{"loss": 0.6343, "learning_rate": 0.00013175893951208642, "epoch": 1.12, "step": 30320}
+{"loss": 0.6885, "learning_rate": 0.00013170324161746683, "epoch": 1.12, "step": 30330}
+{"loss": 0.5956, "learning_rate": 0.00013164754372284727, "epoch": 1.12, "step": 30340}
+{"loss": 0.5588, "learning_rate": 0.00013159184582822768, "epoch": 1.12, "step": 30350}
+{"loss": 0.5711, "learning_rate": 0.0001315361479336081, "epoch": 1.13, "step": 30360}
+{"loss": 0.6897, "learning_rate": 0.00013148045003898853, "epoch": 1.13, "step": 30370}
+{"loss": 0.7197, "learning_rate": 0.00013142475214436894, "epoch": 1.13, "step": 30380}
+{"loss": 0.6109, "learning_rate": 0.00013136905424974935, "epoch": 1.13, "step": 30390}
+{"loss": 0.648, "learning_rate": 0.00013131335635512977, "epoch": 1.13, "step": 30400}
+{"loss": 0.5518, "learning_rate": 0.00013125765846051018, "epoch": 1.13, "step": 30410}
+{"loss": 0.8319, "learning_rate": 0.0001312019605658906, "epoch": 1.13, "step": 30420}
+{"loss": 0.5594, "learning_rate": 0.000131146262671271, "epoch": 1.13, "step": 30430}
+{"loss": 0.583, "learning_rate": 0.00013109056477665144, "epoch": 1.13, "step": 30440}
+{"loss": 0.6148, "learning_rate": 0.00013103486688203185, "epoch": 1.13, "step": 30450}
+{"loss": 0.5644, "learning_rate": 0.00013097916898741226, "epoch": 1.13, "step": 30460}
+{"loss": 0.4781, "learning_rate": 0.0001309234710927927, "epoch": 1.13, "step": 30470}
+{"loss": 0.6261, "learning_rate": 0.0001308677731981731, "epoch": 1.13, "step": 30480}
+{"loss": 0.6081, "learning_rate": 0.00013081207530355352, "epoch": 1.13, "step": 30490}
+{"loss": 0.7234, "learning_rate": 0.00013075637740893393, "epoch": 1.13, "step": 30500}
+{"loss": 0.5754, "learning_rate": 0.00013070067951431434, "epoch": 1.13, "step": 30510}
+{"loss": 0.7789, "learning_rate": 0.00013064498161969475, "epoch": 1.13, "step": 30520}
+{"loss": 0.6051, "learning_rate": 0.0001305892837250752, "epoch": 1.13, "step": 30530}
+{"loss": 0.5934, "learning_rate": 0.0001305335858304556, "epoch": 1.13, "step": 30540}
+{"loss": 0.6367, "learning_rate": 0.00013047788793583601, "epoch": 1.13, "step": 30550}
+{"loss": 0.6543, "learning_rate": 0.00013042219004121643, "epoch": 1.13, "step": 30560}
+{"loss": 0.6487, "learning_rate": 0.00013036649214659686, "epoch": 1.13, "step": 30570}
+{"loss": 0.6244, "learning_rate": 0.00013031079425197727, "epoch": 1.13, "step": 30580}
+{"loss": 0.516, "learning_rate": 0.00013025509635735769, "epoch": 1.13, "step": 30590}
+{"loss": 0.506, "learning_rate": 0.0001301993984627381, "epoch": 1.13, "step": 30600}
+{"loss": 0.7163, "learning_rate": 0.0001301437005681185, "epoch": 1.13, "step": 30610}
+{"loss": 0.5847, "learning_rate": 0.00013008800267349892, "epoch": 1.13, "step": 30620}
+{"loss": 0.5561, "learning_rate": 0.00013003230477887936, "epoch": 1.14, "step": 30630}
+{"loss": 0.5224, "learning_rate": 0.00012997660688425977, "epoch": 1.14, "step": 30640}
+{"loss": 0.5045, "learning_rate": 0.00012992090898964018, "epoch": 1.14, "step": 30650}
+{"loss": 0.6612, "learning_rate": 0.00012986521109502062, "epoch": 1.14, "step": 30660}
+{"loss": 0.6424, "learning_rate": 0.00012980951320040103, "epoch": 1.14, "step": 30670}
+{"loss": 0.5195, "learning_rate": 0.00012975381530578144, "epoch": 1.14, "step": 30680}
+{"loss": 0.5494, "learning_rate": 0.00012969811741116185, "epoch": 1.14, "step": 30690}
+{"loss": 0.6533, "learning_rate": 0.00012964241951654226, "epoch": 1.14, "step": 30700}
+{"loss": 0.6205, "learning_rate": 0.00012958672162192267, "epoch": 1.14, "step": 30710}
+{"loss": 0.5592, "learning_rate": 0.00012953102372730309, "epoch": 1.14, "step": 30720}
+{"loss": 0.5923, "learning_rate": 0.00012947532583268352, "epoch": 1.14, "step": 30730}
+{"loss": 0.5933, "learning_rate": 0.00012941962793806393, "epoch": 1.14, "step": 30740}
+{"loss": 0.6032, "learning_rate": 0.00012936393004344435, "epoch": 1.14, "step": 30750}
+{"loss": 0.5956, "learning_rate": 0.00012930823214882476, "epoch": 1.14, "step": 30760}
+{"loss": 0.6083, "learning_rate": 0.00012925253425420517, "epoch": 1.14, "step": 30770}
+{"loss": 0.5612, "learning_rate": 0.0001291968363595856, "epoch": 1.14, "step": 30780}
+{"loss": 0.6401, "learning_rate": 0.00012914113846496602, "epoch": 1.14, "step": 30790}
+{"loss": 0.5297, "learning_rate": 0.00012908544057034643, "epoch": 1.14, "step": 30800}
+{"loss": 0.6286, "learning_rate": 0.00012902974267572684, "epoch": 1.14, "step": 30810}
+{"loss": 0.5737, "learning_rate": 0.00012897404478110725, "epoch": 1.14, "step": 30820}
+{"loss": 0.5317, "learning_rate": 0.0001289183468864877, "epoch": 1.14, "step": 30830}
+{"loss": 0.5812, "learning_rate": 0.0001288626489918681, "epoch": 1.14, "step": 30840}
+{"loss": 0.6965, "learning_rate": 0.0001288069510972485, "epoch": 1.14, "step": 30850}
+{"loss": 0.5091, "learning_rate": 0.00012875125320262892, "epoch": 1.14, "step": 30860}
+{"loss": 0.3956, "learning_rate": 0.00012869555530800933, "epoch": 1.14, "step": 30870}
+{"loss": 0.6235, "learning_rate": 0.00012863985741338974, "epoch": 1.14, "step": 30880}
+{"loss": 0.7077, "learning_rate": 0.00012858415951877018, "epoch": 1.14, "step": 30890}
+{"loss": 0.6143, "learning_rate": 0.0001285284616241506, "epoch": 1.15, "step": 30900}
+{"loss": 0.5636, "learning_rate": 0.000128472763729531, "epoch": 1.15, "step": 30910}
+{"loss": 0.6609, "learning_rate": 0.00012841706583491144, "epoch": 1.15, "step": 30920}
+{"loss": 0.4568, "learning_rate": 0.00012836136794029185, "epoch": 1.15, "step": 30930}
+{"loss": 0.5697, "learning_rate": 0.00012830567004567227, "epoch": 1.15, "step": 30940}
+{"loss": 0.557, "learning_rate": 0.00012824997215105268, "epoch": 1.15, "step": 30950}
+{"loss": 0.5489, "learning_rate": 0.0001281942742564331, "epoch": 1.15, "step": 30960}
+{"loss": 0.6089, "learning_rate": 0.0001281385763618135, "epoch": 1.15, "step": 30970}
+{"loss": 0.6368, "learning_rate": 0.0001280828784671939, "epoch": 1.15, "step": 30980}
+{"loss": 0.6284, "learning_rate": 0.00012802718057257435, "epoch": 1.15, "step": 30990}
+{"loss": 0.6526, "learning_rate": 0.00012797148267795476, "epoch": 1.15, "step": 31000}
+{"loss": 0.5384, "learning_rate": 0.00012791578478333517, "epoch": 1.15, "step": 31010}
+{"loss": 0.5488, "learning_rate": 0.0001278600868887156, "epoch": 1.15, "step": 31020}
+{"loss": 0.5408, "learning_rate": 0.00012780438899409602, "epoch": 1.15, "step": 31030}
+{"loss": 0.6549, "learning_rate": 0.00012774869109947643, "epoch": 1.15, "step": 31040}
+{"loss": 0.6361, "learning_rate": 0.00012769299320485684, "epoch": 1.15, "step": 31050}
+{"loss": 0.6188, "learning_rate": 0.00012763729531023725, "epoch": 1.15, "step": 31060}
+{"loss": 0.6195, "learning_rate": 0.00012758159741561766, "epoch": 1.15, "step": 31070}
+{"loss": 0.6442, "learning_rate": 0.00012752589952099808, "epoch": 1.15, "step": 31080}
+{"loss": 0.6078, "learning_rate": 0.00012747020162637851, "epoch": 1.15, "step": 31090}
+{"loss": 0.6378, "learning_rate": 0.00012741450373175893, "epoch": 1.15, "step": 31100}
+{"loss": 0.6, "learning_rate": 0.00012735880583713934, "epoch": 1.15, "step": 31110}
+{"loss": 0.4613, "learning_rate": 0.00012730310794251977, "epoch": 1.15, "step": 31120}
+{"loss": 0.5122, "learning_rate": 0.00012724741004790019, "epoch": 1.15, "step": 31130}
+{"loss": 0.7254, "learning_rate": 0.0001271917121532806, "epoch": 1.15, "step": 31140}
+{"loss": 0.5964, "learning_rate": 0.000127136014258661, "epoch": 1.15, "step": 31150}
+{"loss": 0.6135, "learning_rate": 0.00012708031636404142, "epoch": 1.15, "step": 31160}
+{"loss": 0.5749, "learning_rate": 0.00012702461846942183, "epoch": 1.16, "step": 31170}
+{"loss": 0.6659, "learning_rate": 0.00012696892057480227, "epoch": 1.16, "step": 31180}
+{"loss": 0.7007, "learning_rate": 0.00012691322268018268, "epoch": 1.16, "step": 31190}
+{"loss": 0.6446, "learning_rate": 0.0001268575247855631, "epoch": 1.16, "step": 31200}
+{"loss": 0.6617, "learning_rate": 0.0001268018268909435, "epoch": 1.16, "step": 31210}
+{"loss": 0.657, "learning_rate": 0.00012674612899632394, "epoch": 1.16, "step": 31220}
+{"loss": 0.5117, "learning_rate": 0.00012669043110170435, "epoch": 1.16, "step": 31230}
+{"loss": 0.5951, "learning_rate": 0.00012663473320708476, "epoch": 1.16, "step": 31240}
+{"loss": 0.5559, "learning_rate": 0.00012657903531246517, "epoch": 1.16, "step": 31250}
+{"loss": 0.6683, "learning_rate": 0.00012652333741784558, "epoch": 1.16, "step": 31260}
+{"loss": 0.5747, "learning_rate": 0.000126467639523226, "epoch": 1.16, "step": 31270}
+{"loss": 0.6786, "learning_rate": 0.00012641194162860643, "epoch": 1.16, "step": 31280}
+{"loss": 0.5221, "learning_rate": 0.00012635624373398685, "epoch": 1.16, "step": 31290}
+{"loss": 0.5608, "learning_rate": 0.00012630054583936726, "epoch": 1.16, "step": 31300}
+{"loss": 0.4458, "learning_rate": 0.0001262448479447477, "epoch": 1.16, "step": 31310}
+{"loss": 0.6528, "learning_rate": 0.0001261891500501281, "epoch": 1.16, "step": 31320}
+{"loss": 0.5739, "learning_rate": 0.00012613345215550852, "epoch": 1.16, "step": 31330}
+{"loss": 0.7336, "learning_rate": 0.00012607775426088893, "epoch": 1.16, "step": 31340}
+{"loss": 0.5657, "learning_rate": 0.00012602205636626934, "epoch": 1.16, "step": 31350}
+{"loss": 0.6638, "learning_rate": 0.00012596635847164975, "epoch": 1.16, "step": 31360}
+{"loss": 0.6651, "learning_rate": 0.00012591066057703016, "epoch": 1.16, "step": 31370}
+{"loss": 0.4942, "learning_rate": 0.0001258549626824106, "epoch": 1.16, "step": 31380}
+{"loss": 0.4958, "learning_rate": 0.000125799264787791, "epoch": 1.16, "step": 31390}
+{"loss": 0.6147, "learning_rate": 0.00012574356689317142, "epoch": 1.16, "step": 31400}
+{"loss": 0.5585, "learning_rate": 0.00012568786899855186, "epoch": 1.16, "step": 31410}
+{"loss": 0.5992, "learning_rate": 0.00012563217110393227, "epoch": 1.16, "step": 31420}
+{"loss": 0.4611, "learning_rate": 0.00012557647320931268, "epoch": 1.16, "step": 31430}
+{"loss": 0.5153, "learning_rate": 0.0001255207753146931, "epoch": 1.17, "step": 31440}
+{"loss": 0.5479, "learning_rate": 0.0001254650774200735, "epoch": 1.17, "step": 31450}
+{"loss": 0.6117, "learning_rate": 0.00012540937952545392, "epoch": 1.17, "step": 31460}
+{"loss": 0.5487, "learning_rate": 0.00012535368163083435, "epoch": 1.17, "step": 31470}
+{"loss": 0.675, "learning_rate": 0.00012529798373621477, "epoch": 1.17, "step": 31480}
+{"loss": 0.6062, "learning_rate": 0.00012524228584159518, "epoch": 1.17, "step": 31490}
+{"loss": 0.5955, "learning_rate": 0.0001251865879469756, "epoch": 1.17, "step": 31500}
+{"loss": 0.6006, "learning_rate": 0.00012513089005235603, "epoch": 1.17, "step": 31510}
+{"loss": 0.5306, "learning_rate": 0.00012507519215773644, "epoch": 1.17, "step": 31520}
+{"loss": 0.621, "learning_rate": 0.00012501949426311685, "epoch": 1.17, "step": 31530}
+{"loss": 0.5824, "learning_rate": 0.00012496379636849726, "epoch": 1.17, "step": 31540}
+{"loss": 0.5662, "learning_rate": 0.00012490809847387767, "epoch": 1.17, "step": 31550}
+{"loss": 0.6175, "learning_rate": 0.00012485240057925808, "epoch": 1.17, "step": 31560}
+{"loss": 0.6209, "learning_rate": 0.00012479670268463852, "epoch": 1.17, "step": 31570}
+{"loss": 0.465, "learning_rate": 0.00012474100479001893, "epoch": 1.17, "step": 31580}
+{"loss": 0.6752, "learning_rate": 0.00012468530689539934, "epoch": 1.17, "step": 31590}
+{"loss": 0.6325, "learning_rate": 0.00012462960900077978, "epoch": 1.17, "step": 31600}
+{"loss": 0.6166, "learning_rate": 0.0001245739111061602, "epoch": 1.17, "step": 31610}
+{"loss": 0.6125, "learning_rate": 0.0001245182132115406, "epoch": 1.17, "step": 31620}
+{"loss": 0.6336, "learning_rate": 0.000124462515316921, "epoch": 1.17, "step": 31630}
+{"loss": 0.6797, "learning_rate": 0.00012440681742230142, "epoch": 1.17, "step": 31640}
+{"loss": 0.5689, "learning_rate": 0.00012435111952768184, "epoch": 1.17, "step": 31650}
+{"loss": 0.7314, "learning_rate": 0.00012429542163306225, "epoch": 1.17, "step": 31660}
+{"loss": 0.5623, "learning_rate": 0.00012423972373844269, "epoch": 1.17, "step": 31670}
+{"loss": 0.599, "learning_rate": 0.0001241840258438231, "epoch": 1.17, "step": 31680}
+{"loss": 0.5516, "learning_rate": 0.0001241283279492035, "epoch": 1.17, "step": 31690}
+{"loss": 0.7278, "learning_rate": 0.00012407263005458395, "epoch": 1.17, "step": 31700}
+{"loss": 0.6036, "learning_rate": 0.00012401693215996436, "epoch": 1.18, "step": 31710}
+{"loss": 0.6737, "learning_rate": 0.00012396123426534477, "epoch": 1.18, "step": 31720}
+{"loss": 0.6582, "learning_rate": 0.00012390553637072518, "epoch": 1.18, "step": 31730}
+{"loss": 0.615, "learning_rate": 0.0001238498384761056, "epoch": 1.18, "step": 31740}
+{"loss": 0.5693, "learning_rate": 0.000123794140581486, "epoch": 1.18, "step": 31750}
+{"loss": 0.6386, "learning_rate": 0.0001237384426868664, "epoch": 1.18, "step": 31760}
+{"loss": 0.7385, "learning_rate": 0.00012368274479224685, "epoch": 1.18, "step": 31770}
+{"loss": 0.4999, "learning_rate": 0.00012362704689762726, "epoch": 1.18, "step": 31780}
+{"loss": 0.5856, "learning_rate": 0.00012357134900300767, "epoch": 1.18, "step": 31790}
+{"loss": 0.5285, "learning_rate": 0.00012351565110838808, "epoch": 1.18, "step": 31800}
+{"loss": 0.5889, "learning_rate": 0.00012345995321376852, "epoch": 1.18, "step": 31810}
+{"loss": 0.5239, "learning_rate": 0.00012340425531914893, "epoch": 1.18, "step": 31820}
+{"loss": 0.5242, "learning_rate": 0.00012334855742452934, "epoch": 1.18, "step": 31830}
+{"loss": 0.5063, "learning_rate": 0.00012329285952990976, "epoch": 1.18, "step": 31840}
+{"loss": 0.568, "learning_rate": 0.00012323716163529017, "epoch": 1.18, "step": 31850}
+{"loss": 0.5067, "learning_rate": 0.0001231814637406706, "epoch": 1.18, "step": 31860}
+{"loss": 0.6699, "learning_rate": 0.00012312576584605102, "epoch": 1.18, "step": 31870}
+{"loss": 0.6278, "learning_rate": 0.00012307006795143143, "epoch": 1.18, "step": 31880}
+{"loss": 0.6087, "learning_rate": 0.00012301437005681184, "epoch": 1.18, "step": 31890}
+{"loss": 0.5501, "learning_rate": 0.00012295867216219225, "epoch": 1.18, "step": 31900}
+{"loss": 0.5187, "learning_rate": 0.00012290297426757266, "epoch": 1.18, "step": 31910}
+{"loss": 0.6098, "learning_rate": 0.00012284727637295307, "epoch": 1.18, "step": 31920}
+{"loss": 0.5598, "learning_rate": 0.0001227915784783335, "epoch": 1.18, "step": 31930}
+{"loss": 0.6283, "learning_rate": 0.00012273588058371392, "epoch": 1.18, "step": 31940}
+{"loss": 0.5931, "learning_rate": 0.00012268018268909433, "epoch": 1.18, "step": 31950}
+{"loss": 0.6998, "learning_rate": 0.00012262448479447477, "epoch": 1.18, "step": 31960}
+{"loss": 0.6068, "learning_rate": 0.00012256878689985518, "epoch": 1.18, "step": 31970}
+{"loss": 0.6025, "learning_rate": 0.0001225130890052356, "epoch": 1.19, "step": 31980}
+{"loss": 0.4986, "learning_rate": 0.000122457391110616, "epoch": 1.19, "step": 31990}
+{"loss": 0.5281, "learning_rate": 0.00012240169321599642, "epoch": 1.19, "step": 32000}
+{"loss": 0.4619, "learning_rate": 0.00012234599532137683, "epoch": 1.19, "step": 32010}
+{"loss": 0.5052, "learning_rate": 0.00012229029742675724, "epoch": 1.19, "step": 32020}
+{"loss": 0.6326, "learning_rate": 0.00012223459953213768, "epoch": 1.19, "step": 32030}
+{"loss": 0.6529, "learning_rate": 0.0001221789016375181, "epoch": 1.19, "step": 32040}
+{"loss": 0.5777, "learning_rate": 0.0001221232037428985, "epoch": 1.19, "step": 32050}
+{"loss": 0.4897, "learning_rate": 0.00012206750584827892, "epoch": 1.19, "step": 32060}
+{"loss": 0.6097, "learning_rate": 0.00012201180795365935, "epoch": 1.19, "step": 32070}
+{"loss": 0.6802, "learning_rate": 0.00012195611005903976, "epoch": 1.19, "step": 32080}
+{"loss": 0.6042, "learning_rate": 0.00012190041216442018, "epoch": 1.19, "step": 32090}
+{"loss": 0.4821, "learning_rate": 0.0001218447142698006, "epoch": 1.19, "step": 32100}
+{"loss": 0.7123, "learning_rate": 0.000121789016375181, "epoch": 1.19, "step": 32110}
+{"loss": 0.6211, "learning_rate": 0.00012173331848056143, "epoch": 1.19, "step": 32120}
+{"loss": 0.5428, "learning_rate": 0.00012167762058594184, "epoch": 1.19, "step": 32130}
+{"loss": 0.5707, "learning_rate": 0.00012162192269132225, "epoch": 1.19, "step": 32140}
+{"loss": 0.585, "learning_rate": 0.00012156622479670266, "epoch": 1.19, "step": 32150}
+{"loss": 0.5349, "learning_rate": 0.00012151052690208309, "epoch": 1.19, "step": 32160}
+{"loss": 0.5231, "learning_rate": 0.0001214548290074635, "epoch": 1.19, "step": 32170}
+{"loss": 0.6829, "learning_rate": 0.00012139913111284391, "epoch": 1.19, "step": 32180}
+{"loss": 0.5766, "learning_rate": 0.00012134343321822435, "epoch": 1.19, "step": 32190}
+{"loss": 0.5331, "learning_rate": 0.00012128773532360476, "epoch": 1.19, "step": 32200}
+{"loss": 0.6083, "learning_rate": 0.00012123203742898517, "epoch": 1.19, "step": 32210}
+{"loss": 0.6622, "learning_rate": 0.0001211763395343656, "epoch": 1.19, "step": 32220}
+{"loss": 0.657, "learning_rate": 0.00012112064163974601, "epoch": 1.19, "step": 32230}
+{"loss": 0.556, "learning_rate": 0.00012106494374512642, "epoch": 1.19, "step": 32240}
+{"loss": 0.6711, "learning_rate": 0.00012100924585050684, "epoch": 1.2, "step": 32250}
+{"loss": 0.6491, "learning_rate": 0.00012095354795588725, "epoch": 1.2, "step": 32260}
+{"loss": 0.6109, "learning_rate": 0.00012089785006126766, "epoch": 1.2, "step": 32270}
+{"loss": 0.6375, "learning_rate": 0.00012084215216664808, "epoch": 1.2, "step": 32280}
+{"loss": 0.5367, "learning_rate": 0.00012078645427202851, "epoch": 1.2, "step": 32290}
+{"loss": 0.4988, "learning_rate": 0.00012073075637740893, "epoch": 1.2, "step": 32300}
+{"loss": 0.5967, "learning_rate": 0.00012067505848278934, "epoch": 1.2, "step": 32310}
+{"loss": 0.8054, "learning_rate": 0.00012061936058816976, "epoch": 1.2, "step": 32320}
+{"loss": 0.6418, "learning_rate": 0.00012056366269355017, "epoch": 1.2, "step": 32330}
+{"loss": 0.5394, "learning_rate": 0.00012050796479893058, "epoch": 1.2, "step": 32340}
+{"loss": 0.532, "learning_rate": 0.00012045226690431101, "epoch": 1.2, "step": 32350}
+{"loss": 0.5286, "learning_rate": 0.00012039656900969142, "epoch": 1.2, "step": 32360}
+{"loss": 0.4502, "learning_rate": 0.00012034087111507183, "epoch": 1.2, "step": 32370}
+{"loss": 0.5606, "learning_rate": 0.00012028517322045227, "epoch": 1.2, "step": 32380}
+{"loss": 0.6236, "learning_rate": 0.00012022947532583268, "epoch": 1.2, "step": 32390}
+{"loss": 0.5886, "learning_rate": 0.00012017377743121309, "epoch": 1.2, "step": 32400}
+{"loss": 0.6355, "learning_rate": 0.00012011807953659352, "epoch": 1.2, "step": 32410}
+{"loss": 0.7605, "learning_rate": 0.00012006238164197393, "epoch": 1.2, "step": 32420}
+{"loss": 0.6216, "learning_rate": 0.00012000668374735434, "epoch": 1.2, "step": 32430}
+{"loss": 0.5012, "learning_rate": 0.00011995098585273475, "epoch": 1.2, "step": 32440}
+{"loss": 0.5292, "learning_rate": 0.00011989528795811517, "epoch": 1.2, "step": 32450}
+{"loss": 0.5439, "learning_rate": 0.00011983959006349558, "epoch": 1.2, "step": 32460}
+{"loss": 0.5853, "learning_rate": 0.000119783892168876, "epoch": 1.2, "step": 32470}
+{"loss": 0.5336, "learning_rate": 0.00011972819427425643, "epoch": 1.2, "step": 32480}
+{"loss": 0.6061, "learning_rate": 0.00011967249637963685, "epoch": 1.2, "step": 32490}
+{"loss": 0.6228, "learning_rate": 0.00011961679848501726, "epoch": 1.2, "step": 32500}
+{"loss": 0.6121, "learning_rate": 0.00011956667037985963, "epoch": 1.2, "step": 32510}
+{"loss": 0.8102, "learning_rate": 0.00011951097248524005, "epoch": 1.21, "step": 32520}
+{"loss": 0.6016, "learning_rate": 0.00011945527459062047, "epoch": 1.21, "step": 32530}
+{"loss": 0.6931, "learning_rate": 0.00011939957669600088, "epoch": 1.21, "step": 32540}
+{"loss": 0.603, "learning_rate": 0.0001193438788013813, "epoch": 1.21, "step": 32550}
+{"loss": 0.5655, "learning_rate": 0.00011928818090676171, "epoch": 1.21, "step": 32560}
+{"loss": 0.7121, "learning_rate": 0.00011923248301214212, "epoch": 1.21, "step": 32570}
+{"loss": 0.6758, "learning_rate": 0.00011917678511752256, "epoch": 1.21, "step": 32580}
+{"loss": 0.5436, "learning_rate": 0.00011912108722290297, "epoch": 1.21, "step": 32590}
+{"loss": 0.6406, "learning_rate": 0.00011906538932828338, "epoch": 1.21, "step": 32600}
+{"loss": 0.4377, "learning_rate": 0.0001190096914336638, "epoch": 1.21, "step": 32610}
+{"loss": 0.5245, "learning_rate": 0.00011895399353904422, "epoch": 1.21, "step": 32620}
+{"loss": 0.5301, "learning_rate": 0.00011889829564442463, "epoch": 1.21, "step": 32630}
+{"loss": 0.5823, "learning_rate": 0.00011884259774980504, "epoch": 1.21, "step": 32640}
+{"loss": 0.5917, "learning_rate": 0.00011878689985518547, "epoch": 1.21, "step": 32650}
+{"loss": 0.5153, "learning_rate": 0.00011873120196056588, "epoch": 1.21, "step": 32660}
+{"loss": 0.5509, "learning_rate": 0.00011867550406594629, "epoch": 1.21, "step": 32670}
+{"loss": 0.5931, "learning_rate": 0.00011861980617132671, "epoch": 1.21, "step": 32680}
+{"loss": 0.6181, "learning_rate": 0.00011856410827670714, "epoch": 1.21, "step": 32690}
+{"loss": 0.4056, "learning_rate": 0.00011850841038208755, "epoch": 1.21, "step": 32700}
+{"loss": 0.5899, "learning_rate": 0.00011845271248746797, "epoch": 1.21, "step": 32710}
+{"loss": 0.5242, "learning_rate": 0.00011839701459284839, "epoch": 1.21, "step": 32720}
+{"loss": 0.7591, "learning_rate": 0.0001183413166982288, "epoch": 1.21, "step": 32730}
+{"loss": 0.7375, "learning_rate": 0.00011828561880360921, "epoch": 1.21, "step": 32740}
+{"loss": 0.5444, "learning_rate": 0.00011822992090898963, "epoch": 1.21, "step": 32750}
+{"loss": 0.5866, "learning_rate": 0.00011817422301437004, "epoch": 1.21, "step": 32760}
+{"loss": 0.6804, "learning_rate": 0.00011811852511975045, "epoch": 1.21, "step": 32770}
+{"loss": 0.5704, "learning_rate": 0.00011806282722513088, "epoch": 1.21, "step": 32780}
+{"loss": 0.7149, "learning_rate": 0.00011800712933051129, "epoch": 1.22, "step": 32790}
+{"loss": 0.565, "learning_rate": 0.0001179514314358917, "epoch": 1.22, "step": 32800}
+{"loss": 0.5827, "learning_rate": 0.00011789573354127214, "epoch": 1.22, "step": 32810}
+{"loss": 0.483, "learning_rate": 0.00011784003564665255, "epoch": 1.22, "step": 32820}
+{"loss": 0.5689, "learning_rate": 0.00011778433775203296, "epoch": 1.22, "step": 32830}
+{"loss": 0.5791, "learning_rate": 0.00011772863985741339, "epoch": 1.22, "step": 32840}
+{"loss": 0.5717, "learning_rate": 0.0001176729419627938, "epoch": 1.22, "step": 32850}
+{"loss": 0.5968, "learning_rate": 0.00011761724406817421, "epoch": 1.22, "step": 32860}
+{"loss": 0.6956, "learning_rate": 0.00011756154617355462, "epoch": 1.22, "step": 32870}
+{"loss": 0.7132, "learning_rate": 0.00011750584827893504, "epoch": 1.22, "step": 32880}
+{"loss": 0.5122, "learning_rate": 0.00011745015038431546, "epoch": 1.22, "step": 32890}
+{"loss": 0.5693, "learning_rate": 0.00011739445248969587, "epoch": 1.22, "step": 32900}
+{"loss": 0.5706, "learning_rate": 0.0001173387545950763, "epoch": 1.22, "step": 32910}
+{"loss": 0.5681, "learning_rate": 0.00011728305670045672, "epoch": 1.22, "step": 32920}
+{"loss": 0.649, "learning_rate": 0.00011722735880583713, "epoch": 1.22, "step": 32930}
+{"loss": 0.745, "learning_rate": 0.00011717166091121755, "epoch": 1.22, "step": 32940}
+{"loss": 0.5371, "learning_rate": 0.00011711596301659796, "epoch": 1.22, "step": 32950}
+{"loss": 0.6556, "learning_rate": 0.00011706026512197837, "epoch": 1.22, "step": 32960}
+{"loss": 0.6298, "learning_rate": 0.0001170045672273588, "epoch": 1.22, "step": 32970}
+{"loss": 0.6426, "learning_rate": 0.00011694886933273921, "epoch": 1.22, "step": 32980}
+{"loss": 0.5628, "learning_rate": 0.00011689317143811962, "epoch": 1.22, "step": 32990}
+{"loss": 0.5488, "learning_rate": 0.00011683747354350003, "epoch": 1.22, "step": 33000}
+{"loss": 0.7095, "learning_rate": 0.00011678177564888047, "epoch": 1.22, "step": 33010}
+{"loss": 0.5199, "learning_rate": 0.00011672607775426088, "epoch": 1.22, "step": 33020}
+{"loss": 0.5917, "learning_rate": 0.00011667037985964129, "epoch": 1.22, "step": 33030}
+{"loss": 0.5515, "learning_rate": 0.00011661468196502172, "epoch": 1.22, "step": 33040}
+{"loss": 0.5361, "learning_rate": 0.00011655898407040213, "epoch": 1.22, "step": 33050}
+{"loss": 0.6972, "learning_rate": 0.00011650328617578254, "epoch": 1.23, "step": 33060}
+{"loss": 0.5692, "learning_rate": 0.00011644758828116296, "epoch": 1.23, "step": 33070}
+{"loss": 0.6353, "learning_rate": 0.00011639189038654338, "epoch": 1.23, "step": 33080}
+{"loss": 0.6797, "learning_rate": 0.00011633619249192379, "epoch": 1.23, "step": 33090}
+{"loss": 0.5158, "learning_rate": 0.00011628049459730423, "epoch": 1.23, "step": 33100}
+{"loss": 0.7578, "learning_rate": 0.00011622479670268464, "epoch": 1.23, "step": 33110}
+{"loss": 0.6854, "learning_rate": 0.00011616909880806505, "epoch": 1.23, "step": 33120}
+{"loss": 0.5312, "learning_rate": 0.00011611340091344546, "epoch": 1.23, "step": 33130}
+{"loss": 0.602, "learning_rate": 0.00011605770301882588, "epoch": 1.23, "step": 33140}
+{"loss": 0.5676, "learning_rate": 0.0001160020051242063, "epoch": 1.23, "step": 33150}
+{"loss": 0.5214, "learning_rate": 0.0001159463072295867, "epoch": 1.23, "step": 33160}
+{"loss": 0.5496, "learning_rate": 0.00011589060933496713, "epoch": 1.23, "step": 33170}
+{"loss": 0.5627, "learning_rate": 0.00011583491144034754, "epoch": 1.23, "step": 33180}
+{"loss": 0.4726, "learning_rate": 0.00011577921354572795, "epoch": 1.23, "step": 33190}
+{"loss": 0.5614, "learning_rate": 0.00011572351565110838, "epoch": 1.23, "step": 33200}
+{"loss": 0.7228, "learning_rate": 0.0001156678177564888, "epoch": 1.23, "step": 33210}
+{"loss": 0.7056, "learning_rate": 0.00011561211986186921, "epoch": 1.23, "step": 33220}
+{"loss": 0.6883, "learning_rate": 0.00011555642196724964, "epoch": 1.23, "step": 33230}
+{"loss": 0.7162, "learning_rate": 0.00011550072407263005, "epoch": 1.23, "step": 33240}
+{"loss": 0.5727, "learning_rate": 0.00011544502617801046, "epoch": 1.23, "step": 33250}
+{"loss": 0.5708, "learning_rate": 0.00011538932828339087, "epoch": 1.23, "step": 33260}
+{"loss": 0.53, "learning_rate": 0.0001153336303887713, "epoch": 1.23, "step": 33270}
+{"loss": 0.5894, "learning_rate": 0.00011527793249415171, "epoch": 1.23, "step": 33280}
+{"loss": 0.5931, "learning_rate": 0.00011522223459953212, "epoch": 1.23, "step": 33290}
+{"loss": 0.5561, "learning_rate": 0.00011516653670491254, "epoch": 1.23, "step": 33300}
+{"loss": 0.5732, "learning_rate": 0.00011511083881029295, "epoch": 1.23, "step": 33310}
+{"loss": 0.7054, "learning_rate": 0.00011505514091567337, "epoch": 1.23, "step": 33320}
+{"loss": 0.6209, "learning_rate": 0.0001149994430210538, "epoch": 1.24, "step": 33330}
+{"loss": 0.5694, "learning_rate": 0.00011494374512643421, "epoch": 1.24, "step": 33340}
+{"loss": 0.49, "learning_rate": 0.00011488804723181463, "epoch": 1.24, "step": 33350}
+{"loss": 0.6434, "learning_rate": 0.00011483234933719505, "epoch": 1.24, "step": 33360}
+{"loss": 0.5169, "learning_rate": 0.00011477665144257546, "epoch": 1.24, "step": 33370}
+{"loss": 0.6233, "learning_rate": 0.00011472095354795587, "epoch": 1.24, "step": 33380}
+{"loss": 0.5409, "learning_rate": 0.0001146652556533363, "epoch": 1.24, "step": 33390}
+{"loss": 0.5444, "learning_rate": 0.00011460955775871671, "epoch": 1.24, "step": 33400}
+{"loss": 0.5678, "learning_rate": 0.00011455385986409712, "epoch": 1.24, "step": 33410}
+{"loss": 0.6457, "learning_rate": 0.00011449816196947753, "epoch": 1.24, "step": 33420}
+{"loss": 0.507, "learning_rate": 0.00011444246407485797, "epoch": 1.24, "step": 33430}
+{"loss": 0.5409, "learning_rate": 0.00011438676618023838, "epoch": 1.24, "step": 33440}
+{"loss": 0.5237, "learning_rate": 0.00011433106828561879, "epoch": 1.24, "step": 33450}
+{"loss": 0.5237, "learning_rate": 0.00011427537039099922, "epoch": 1.24, "step": 33460}
+{"loss": 0.3855, "learning_rate": 0.00011421967249637963, "epoch": 1.24, "step": 33470}
+{"loss": 0.6786, "learning_rate": 0.00011416397460176004, "epoch": 1.24, "step": 33480}
+{"loss": 0.7491, "learning_rate": 0.00011410827670714046, "epoch": 1.24, "step": 33490}
+{"loss": 0.5322, "learning_rate": 0.00011405257881252087, "epoch": 1.24, "step": 33500}
+{"loss": 0.4671, "learning_rate": 0.00011399688091790129, "epoch": 1.24, "step": 33510}
+{"loss": 0.6247, "learning_rate": 0.00011394118302328172, "epoch": 1.24, "step": 33520}
+{"loss": 0.5359, "learning_rate": 0.00011388548512866213, "epoch": 1.24, "step": 33530}
+{"loss": 0.5205, "learning_rate": 0.00011382978723404255, "epoch": 1.24, "step": 33540}
+{"loss": 0.5354, "learning_rate": 0.00011377408933942296, "epoch": 1.24, "step": 33550}
+{"loss": 0.4421, "learning_rate": 0.00011371839144480338, "epoch": 1.24, "step": 33560}
+{"loss": 0.5371, "learning_rate": 0.00011366269355018379, "epoch": 1.24, "step": 33570}
+{"loss": 0.6531, "learning_rate": 0.0001136069956555642, "epoch": 1.24, "step": 33580}
+{"loss": 0.5245, "learning_rate": 0.00011355129776094463, "epoch": 1.24, "step": 33590}
+{"loss": 0.5616, "learning_rate": 0.00011349559986632504, "epoch": 1.25, "step": 33600}
+{"loss": 0.5483, "learning_rate": 0.00011343990197170545, "epoch": 1.25, "step": 33610}
+{"loss": 0.5441, "learning_rate": 0.00011338420407708589, "epoch": 1.25, "step": 33620}
+{"loss": 0.6472, "learning_rate": 0.0001133285061824663, "epoch": 1.25, "step": 33630}
+{"loss": 0.6369, "learning_rate": 0.00011327280828784671, "epoch": 1.25, "step": 33640}
+{"loss": 0.6239, "learning_rate": 0.00011321711039322714, "epoch": 1.25, "step": 33650}
+{"loss": 0.554, "learning_rate": 0.00011316141249860755, "epoch": 1.25, "step": 33660}
+{"loss": 0.5735, "learning_rate": 0.00011310571460398796, "epoch": 1.25, "step": 33670}
+{"loss": 0.6482, "learning_rate": 0.00011305001670936837, "epoch": 1.25, "step": 33680}
+{"loss": 0.6359, "learning_rate": 0.0001129943188147488, "epoch": 1.25, "step": 33690}
+{"loss": 0.5881, "learning_rate": 0.0001129386209201292, "epoch": 1.25, "step": 33700}
+{"loss": 0.7353, "learning_rate": 0.00011288292302550962, "epoch": 1.25, "step": 33710}
+{"loss": 0.555, "learning_rate": 0.00011282722513089005, "epoch": 1.25, "step": 33720}
+{"loss": 0.6089, "learning_rate": 0.00011277152723627047, "epoch": 1.25, "step": 33730}
+{"loss": 0.6418, "learning_rate": 0.00011271582934165088, "epoch": 1.25, "step": 33740}
+{"loss": 0.6514, "learning_rate": 0.0001126601314470313, "epoch": 1.25, "step": 33750}
+{"loss": 0.5712, "learning_rate": 0.00011260443355241171, "epoch": 1.25, "step": 33760}
+{"loss": 0.4586, "learning_rate": 0.00011254873565779212, "epoch": 1.25, "step": 33770}
+{"loss": 0.554, "learning_rate": 0.00011249303776317255, "epoch": 1.25, "step": 33780}
+{"loss": 0.5715, "learning_rate": 0.00011243733986855296, "epoch": 1.25, "step": 33790}
+{"loss": 0.5781, "learning_rate": 0.00011238164197393337, "epoch": 1.25, "step": 33800}
+{"loss": 0.6765, "learning_rate": 0.00011232594407931378, "epoch": 1.25, "step": 33810}
+{"loss": 0.5847, "learning_rate": 0.0001122702461846942, "epoch": 1.25, "step": 33820}
+{"loss": 0.6121, "learning_rate": 0.00011221454829007462, "epoch": 1.25, "step": 33830}
+{"loss": 0.4199, "learning_rate": 0.00011215885039545504, "epoch": 1.25, "step": 33840}
+{"loss": 0.6374, "learning_rate": 0.00011210315250083547, "epoch": 1.25, "step": 33850}
+{"loss": 0.5438, "learning_rate": 0.00011204745460621588, "epoch": 1.25, "step": 33860}
+{"loss": 0.5745, "learning_rate": 0.00011199175671159629, "epoch": 1.26, "step": 33870}
+{"loss": 0.5881, "learning_rate": 0.00011193605881697671, "epoch": 1.26, "step": 33880}
+{"loss": 0.7153, "learning_rate": 0.00011188036092235713, "epoch": 1.26, "step": 33890}
+{"loss": 0.5935, "learning_rate": 0.00011182466302773754, "epoch": 1.26, "step": 33900}
+{"loss": 0.5815, "learning_rate": 0.00011176896513311796, "epoch": 1.26, "step": 33910}
+{"loss": 0.607, "learning_rate": 0.00011171326723849837, "epoch": 1.26, "step": 33920}
+{"loss": 0.6031, "learning_rate": 0.00011165756934387878, "epoch": 1.26, "step": 33930}
+{"loss": 0.4774, "learning_rate": 0.0001116018714492592, "epoch": 1.26, "step": 33940}
+{"loss": 0.6115, "learning_rate": 0.00011154617355463963, "epoch": 1.26, "step": 33950}
+{"loss": 0.6182, "learning_rate": 0.00011149047566002004, "epoch": 1.26, "step": 33960}
+{"loss": 0.6672, "learning_rate": 0.00011143477776540045, "epoch": 1.26, "step": 33970}
+{"loss": 0.4804, "learning_rate": 0.00011137907987078088, "epoch": 1.26, "step": 33980}
+{"loss": 0.6426, "learning_rate": 0.00011132338197616129, "epoch": 1.26, "step": 33990}
+{"loss": 0.6869, "learning_rate": 0.0001112676840815417, "epoch": 1.26, "step": 34000}
+{"loss": 0.5712, "learning_rate": 0.00011121198618692213, "epoch": 1.26, "step": 34010}
+{"loss": 0.5906, "learning_rate": 0.00011115628829230254, "epoch": 1.26, "step": 34020}
+{"loss": 0.6791, "learning_rate": 0.00011110059039768295, "epoch": 1.26, "step": 34030}
+{"loss": 0.548, "learning_rate": 0.00011104489250306339, "epoch": 1.26, "step": 34040}
+{"loss": 0.4906, "learning_rate": 0.0001109891946084438, "epoch": 1.26, "step": 34050}
+{"loss": 0.5314, "learning_rate": 0.00011093349671382421, "epoch": 1.26, "step": 34060}
+{"loss": 0.649, "learning_rate": 0.00011087779881920462, "epoch": 1.26, "step": 34070}
+{"loss": 0.6003, "learning_rate": 0.00011082210092458505, "epoch": 1.26, "step": 34080}
+{"loss": 0.5759, "learning_rate": 0.00011076640302996546, "epoch": 1.26, "step": 34090}
+{"loss": 0.6643, "learning_rate": 0.00011071070513534587, "epoch": 1.26, "step": 34100}
+{"loss": 0.7104, "learning_rate": 0.00011065500724072629, "epoch": 1.26, "step": 34110}
+{"loss": 0.5609, "learning_rate": 0.0001105993093461067, "epoch": 1.26, "step": 34120}
+{"loss": 0.5502, "learning_rate": 0.00011054361145148711, "epoch": 1.26, "step": 34130}
+{"loss": 0.4918, "learning_rate": 0.00011048791355686755, "epoch": 1.27, "step": 34140}
+{"loss": 0.6105, "learning_rate": 0.00011043221566224796, "epoch": 1.27, "step": 34150}
+{"loss": 0.5179, "learning_rate": 0.00011037651776762837, "epoch": 1.27, "step": 34160}
+{"loss": 0.5122, "learning_rate": 0.0001103208198730088, "epoch": 1.27, "step": 34170}
+{"loss": 0.6892, "learning_rate": 0.00011026512197838921, "epoch": 1.27, "step": 34180}
+{"loss": 0.7243, "learning_rate": 0.00011020942408376962, "epoch": 1.27, "step": 34190}
+{"loss": 0.5811, "learning_rate": 0.00011015372618915003, "epoch": 1.27, "step": 34200}
+{"loss": 0.5849, "learning_rate": 0.00011009802829453046, "epoch": 1.27, "step": 34210}
+{"loss": 0.4927, "learning_rate": 0.00011004233039991087, "epoch": 1.27, "step": 34220}
+{"loss": 0.5549, "learning_rate": 0.00010998663250529128, "epoch": 1.27, "step": 34230}
+{"loss": 0.5977, "learning_rate": 0.00010993093461067172, "epoch": 1.27, "step": 34240}
+{"loss": 0.6311, "learning_rate": 0.00010987523671605213, "epoch": 1.27, "step": 34250}
+{"loss": 0.5196, "learning_rate": 0.00010981953882143254, "epoch": 1.27, "step": 34260}
+{"loss": 0.5122, "learning_rate": 0.00010976384092681297, "epoch": 1.27, "step": 34270}
+{"loss": 0.5519, "learning_rate": 0.00010970814303219338, "epoch": 1.27, "step": 34280}
+{"loss": 0.6774, "learning_rate": 0.00010965244513757379, "epoch": 1.27, "step": 34290}
+{"loss": 0.553, "learning_rate": 0.00010959674724295421, "epoch": 1.27, "step": 34300}
+{"loss": 0.6607, "learning_rate": 0.00010954104934833462, "epoch": 1.27, "step": 34310}
+{"loss": 0.5861, "learning_rate": 0.00010948535145371503, "epoch": 1.27, "step": 34320}
+{"loss": 0.685, "learning_rate": 0.00010942965355909546, "epoch": 1.27, "step": 34330}
+{"loss": 0.6033, "learning_rate": 0.00010937395566447587, "epoch": 1.27, "step": 34340}
+{"loss": 0.6781, "learning_rate": 0.00010931825776985628, "epoch": 1.27, "step": 34350}
+{"loss": 0.7013, "learning_rate": 0.0001092625598752367, "epoch": 1.27, "step": 34360}
+{"loss": 0.5961, "learning_rate": 0.00010920686198061713, "epoch": 1.27, "step": 34370}
+{"loss": 0.489, "learning_rate": 0.00010915116408599754, "epoch": 1.27, "step": 34380}
+{"loss": 0.5715, "learning_rate": 0.00010909546619137795, "epoch": 1.27, "step": 34390}
+{"loss": 0.757, "learning_rate": 0.00010903976829675838, "epoch": 1.27, "step": 34400}
+{"loss": 0.6089, "learning_rate": 0.00010898407040213879, "epoch": 1.28, "step": 34410}
+{"loss": 0.5625, "learning_rate": 0.0001089283725075192, "epoch": 1.28, "step": 34420}
+{"loss": 0.6743, "learning_rate": 0.00010887267461289962, "epoch": 1.28, "step": 34430}
+{"loss": 0.5399, "learning_rate": 0.00010881697671828004, "epoch": 1.28, "step": 34440}
+{"loss": 0.4402, "learning_rate": 0.00010876127882366045, "epoch": 1.28, "step": 34450}
+{"loss": 0.5848, "learning_rate": 0.00010870558092904089, "epoch": 1.28, "step": 34460}
+{"loss": 0.7084, "learning_rate": 0.0001086498830344213, "epoch": 1.28, "step": 34470}
+{"loss": 0.5794, "learning_rate": 0.00010859418513980171, "epoch": 1.28, "step": 34480}
+{"loss": 0.6234, "learning_rate": 0.00010853848724518212, "epoch": 1.28, "step": 34490}
+{"loss": 0.607, "learning_rate": 0.00010848278935056254, "epoch": 1.28, "step": 34500}
+{"loss": 0.6204, "learning_rate": 0.00010842709145594295, "epoch": 1.28, "step": 34510}
+{"loss": 0.5963, "learning_rate": 0.00010837139356132337, "epoch": 1.28, "step": 34520}
+{"loss": 0.5794, "learning_rate": 0.00010831569566670379, "epoch": 1.28, "step": 34530}
+{"loss": 0.4941, "learning_rate": 0.0001082599977720842, "epoch": 1.28, "step": 34540}
+{"loss": 0.5723, "learning_rate": 0.00010820429987746461, "epoch": 1.28, "step": 34550}
+{"loss": 0.5818, "learning_rate": 0.00010814860198284505, "epoch": 1.28, "step": 34560}
+{"loss": 0.6557, "learning_rate": 0.00010809290408822546, "epoch": 1.28, "step": 34570}
+{"loss": 0.6478, "learning_rate": 0.00010803720619360587, "epoch": 1.28, "step": 34580}
+{"loss": 0.4837, "learning_rate": 0.0001079815082989863, "epoch": 1.28, "step": 34590}
+{"loss": 0.5239, "learning_rate": 0.00010792581040436671, "epoch": 1.28, "step": 34600}
+{"loss": 0.503, "learning_rate": 0.00010787011250974712, "epoch": 1.28, "step": 34610}
+{"loss": 0.6065, "learning_rate": 0.00010781441461512753, "epoch": 1.28, "step": 34620}
+{"loss": 0.5867, "learning_rate": 0.00010775871672050796, "epoch": 1.28, "step": 34630}
+{"loss": 0.5584, "learning_rate": 0.00010770301882588837, "epoch": 1.28, "step": 34640}
+{"loss": 0.5843, "learning_rate": 0.00010764732093126878, "epoch": 1.28, "step": 34650}
+{"loss": 0.6004, "learning_rate": 0.00010759162303664922, "epoch": 1.28, "step": 34660}
+{"loss": 0.6147, "learning_rate": 0.00010753592514202963, "epoch": 1.28, "step": 34670}
+{"loss": 0.6261, "learning_rate": 0.00010748022724741004, "epoch": 1.29, "step": 34680}
+{"loss": 0.69, "learning_rate": 0.00010742452935279046, "epoch": 1.29, "step": 34690}
+{"loss": 0.6349, "learning_rate": 0.00010736883145817087, "epoch": 1.29, "step": 34700}
+{"loss": 0.6343, "learning_rate": 0.00010731313356355129, "epoch": 1.29, "step": 34710}
+{"loss": 0.5058, "learning_rate": 0.00010725743566893171, "epoch": 1.29, "step": 34720}
+{"loss": 0.5787, "learning_rate": 0.00010720173777431212, "epoch": 1.29, "step": 34730}
+{"loss": 0.5448, "learning_rate": 0.00010714603987969253, "epoch": 1.29, "step": 34740}
+{"loss": 0.6968, "learning_rate": 0.00010709034198507294, "epoch": 1.29, "step": 34750}
+{"loss": 0.787, "learning_rate": 0.00010703464409045338, "epoch": 1.29, "step": 34760}
+{"loss": 0.5372, "learning_rate": 0.00010697894619583379, "epoch": 1.29, "step": 34770}
+{"loss": 0.5287, "learning_rate": 0.0001069232483012142, "epoch": 1.29, "step": 34780}
+{"loss": 0.6151, "learning_rate": 0.00010686755040659463, "epoch": 1.29, "step": 34790}
+{"loss": 0.7136, "learning_rate": 0.00010681185251197504, "epoch": 1.29, "step": 34800}
+{"loss": 0.6034, "learning_rate": 0.00010675615461735545, "epoch": 1.29, "step": 34810}
+{"loss": 0.5197, "learning_rate": 0.00010670045672273588, "epoch": 1.29, "step": 34820}
+{"loss": 0.5162, "learning_rate": 0.00010664475882811629, "epoch": 1.29, "step": 34830}
+{"loss": 0.5379, "learning_rate": 0.0001065890609334967, "epoch": 1.29, "step": 34840}
+{"loss": 0.5716, "learning_rate": 0.00010653336303887712, "epoch": 1.29, "step": 34850}
+{"loss": 0.7303, "learning_rate": 0.00010647766514425753, "epoch": 1.29, "step": 34860}
+{"loss": 0.6452, "learning_rate": 0.00010642196724963794, "epoch": 1.29, "step": 34870}
+{"loss": 0.5527, "learning_rate": 0.00010636626935501837, "epoch": 1.29, "step": 34880}
+{"loss": 0.5434, "learning_rate": 0.0001063105714603988, "epoch": 1.29, "step": 34890}
+{"loss": 0.6569, "learning_rate": 0.0001062548735657792, "epoch": 1.29, "step": 34900}
+{"loss": 0.6112, "learning_rate": 0.00010619917567115962, "epoch": 1.29, "step": 34910}
+{"loss": 0.5214, "learning_rate": 0.00010614347777654004, "epoch": 1.29, "step": 34920}
+{"loss": 0.5814, "learning_rate": 0.00010608777988192045, "epoch": 1.29, "step": 34930}
+{"loss": 0.5672, "learning_rate": 0.00010603208198730086, "epoch": 1.29, "step": 34940}
+{"loss": 0.673, "learning_rate": 0.00010597638409268129, "epoch": 1.3, "step": 34950}
+{"loss": 0.6287, "learning_rate": 0.0001059206861980617, "epoch": 1.3, "step": 34960}
+{"loss": 0.6001, "learning_rate": 0.00010586498830344211, "epoch": 1.3, "step": 34970}
+{"loss": 0.6409, "learning_rate": 0.00010580929040882255, "epoch": 1.3, "step": 34980}
+{"loss": 0.4455, "learning_rate": 0.00010575359251420296, "epoch": 1.3, "step": 34990}
+{"loss": 0.5373, "learning_rate": 0.00010569789461958337, "epoch": 1.3, "step": 35000}
+{"loss": 0.6713, "learning_rate": 0.00010564219672496378, "epoch": 1.3, "step": 35010}
+{"loss": 0.6506, "learning_rate": 0.00010558649883034421, "epoch": 1.3, "step": 35020}
+{"loss": 0.6333, "learning_rate": 0.00010553080093572462, "epoch": 1.3, "step": 35030}
+{"loss": 0.572, "learning_rate": 0.00010547510304110503, "epoch": 1.3, "step": 35040}
+{"loss": 0.4745, "learning_rate": 0.00010541940514648545, "epoch": 1.3, "step": 35050}
+{"loss": 0.6048, "learning_rate": 0.00010536370725186586, "epoch": 1.3, "step": 35060}
+{"loss": 0.6027, "learning_rate": 0.00010530800935724628, "epoch": 1.3, "step": 35070}
+{"loss": 0.7227, "learning_rate": 0.00010525231146262671, "epoch": 1.3, "step": 35080}
+{"loss": 0.7286, "learning_rate": 0.00010519661356800713, "epoch": 1.3, "step": 35090}
+{"loss": 0.5431, "learning_rate": 0.00010514091567338754, "epoch": 1.3, "step": 35100}
+{"loss": 0.6007, "learning_rate": 0.00010508521777876796, "epoch": 1.3, "step": 35110}
+{"loss": 0.6274, "learning_rate": 0.00010502951988414837, "epoch": 1.3, "step": 35120}
+{"loss": 0.7099, "learning_rate": 0.00010497382198952878, "epoch": 1.3, "step": 35130}
+{"loss": 0.6034, "learning_rate": 0.0001049181240949092, "epoch": 1.3, "step": 35140}
+{"loss": 0.5529, "learning_rate": 0.00010486242620028962, "epoch": 1.3, "step": 35150}
+{"loss": 0.6614, "learning_rate": 0.00010480672830567003, "epoch": 1.3, "step": 35160}
+{"loss": 0.7438, "learning_rate": 0.00010475103041105044, "epoch": 1.3, "step": 35170}
+{"loss": 0.7015, "learning_rate": 0.00010469533251643088, "epoch": 1.3, "step": 35180}
+{"loss": 0.5895, "learning_rate": 0.00010463963462181129, "epoch": 1.3, "step": 35190}
+{"loss": 0.6446, "learning_rate": 0.0001045839367271917, "epoch": 1.3, "step": 35200}
+{"loss": 0.7804, "learning_rate": 0.00010452823883257213, "epoch": 1.3, "step": 35210}
+{"loss": 0.6018, "learning_rate": 0.00010447254093795254, "epoch": 1.31, "step": 35220}
+{"loss": 0.5574, "learning_rate": 0.00010441684304333295, "epoch": 1.31, "step": 35230}
+{"loss": 0.7188, "learning_rate": 0.00010436114514871337, "epoch": 1.31, "step": 35240}
+{"loss": 0.5537, "learning_rate": 0.00010430544725409378, "epoch": 1.31, "step": 35250}
+{"loss": 0.6568, "learning_rate": 0.0001042497493594742, "epoch": 1.31, "step": 35260}
+{"loss": 0.6737, "learning_rate": 0.00010419405146485463, "epoch": 1.31, "step": 35270}
+{"loss": 0.7045, "learning_rate": 0.00010413835357023505, "epoch": 1.31, "step": 35280}
+{"loss": 0.5646, "learning_rate": 0.00010408265567561546, "epoch": 1.31, "step": 35290}
+{"loss": 0.5411, "learning_rate": 0.00010402695778099587, "epoch": 1.31, "step": 35300}
+{"loss": 0.5279, "learning_rate": 0.00010397125988637629, "epoch": 1.31, "step": 35310}
+{"loss": 0.5623, "learning_rate": 0.0001039155619917567, "epoch": 1.31, "step": 35320}
+{"loss": 0.5809, "learning_rate": 0.00010385986409713711, "epoch": 1.31, "step": 35330}
+{"loss": 0.5448, "learning_rate": 0.00010380416620251754, "epoch": 1.31, "step": 35340}
+{"loss": 0.592, "learning_rate": 0.00010374846830789795, "epoch": 1.31, "step": 35350}
+{"loss": 0.6807, "learning_rate": 0.00010369277041327836, "epoch": 1.31, "step": 35360}
+{"loss": 0.5957, "learning_rate": 0.00010363707251865879, "epoch": 1.31, "step": 35370}
+{"loss": 0.5448, "learning_rate": 0.0001035813746240392, "epoch": 1.31, "step": 35380}
+{"loss": 0.5674, "learning_rate": 0.00010352567672941962, "epoch": 1.31, "step": 35390}
+{"loss": 0.6582, "learning_rate": 0.00010346997883480005, "epoch": 1.31, "step": 35400}
+{"loss": 0.5759, "learning_rate": 0.00010341428094018046, "epoch": 1.31, "step": 35410}
+{"loss": 0.678, "learning_rate": 0.00010335858304556087, "epoch": 1.31, "step": 35420}
+{"loss": 0.4513, "learning_rate": 0.00010330288515094128, "epoch": 1.31, "step": 35430}
+{"loss": 0.6136, "learning_rate": 0.0001032471872563217, "epoch": 1.31, "step": 35440}
+{"loss": 0.5059, "learning_rate": 0.00010319148936170212, "epoch": 1.31, "step": 35450}
+{"loss": 0.6887, "learning_rate": 0.00010313579146708253, "epoch": 1.31, "step": 35460}
+{"loss": 0.5988, "learning_rate": 0.00010308009357246295, "epoch": 1.31, "step": 35470}
+{"loss": 0.5616, "learning_rate": 0.00010302439567784336, "epoch": 1.31, "step": 35480}
+{"loss": 0.5207, "learning_rate": 0.00010296869778322377, "epoch": 1.32, "step": 35490}
+{"loss": 0.4487, "learning_rate": 0.00010291299988860421, "epoch": 1.32, "step": 35500}
+{"loss": 0.7288, "learning_rate": 0.00010285730199398462, "epoch": 1.32, "step": 35510}
+{"loss": 0.6065, "learning_rate": 0.00010280160409936503, "epoch": 1.32, "step": 35520}
+{"loss": 0.6433, "learning_rate": 0.00010274590620474546, "epoch": 1.32, "step": 35530}
+{"loss": 0.4878, "learning_rate": 0.00010269020831012587, "epoch": 1.32, "step": 35540}
+{"loss": 0.6387, "learning_rate": 0.00010263451041550628, "epoch": 1.32, "step": 35550}
+{"loss": 0.6496, "learning_rate": 0.00010257881252088669, "epoch": 1.32, "step": 35560}
+{"loss": 0.594, "learning_rate": 0.00010252311462626712, "epoch": 1.32, "step": 35570}
+{"loss": 0.6354, "learning_rate": 0.00010246741673164753, "epoch": 1.32, "step": 35580}
+{"loss": 0.5754, "learning_rate": 0.00010241171883702794, "epoch": 1.32, "step": 35590}
+{"loss": 0.6031, "learning_rate": 0.00010235602094240838, "epoch": 1.32, "step": 35600}
+{"loss": 0.6351, "learning_rate": 0.00010230032304778879, "epoch": 1.32, "step": 35610}
+{"loss": 0.5952, "learning_rate": 0.0001022446251531692, "epoch": 1.32, "step": 35620}
+{"loss": 0.5913, "learning_rate": 0.00010218892725854962, "epoch": 1.32, "step": 35630}
+{"loss": 0.5138, "learning_rate": 0.00010213322936393004, "epoch": 1.32, "step": 35640}
+{"loss": 0.5812, "learning_rate": 0.00010207753146931045, "epoch": 1.32, "step": 35650}
+{"loss": 0.5434, "learning_rate": 0.00010202183357469087, "epoch": 1.32, "step": 35660}
+{"loss": 0.6442, "learning_rate": 0.00010196613568007128, "epoch": 1.32, "step": 35670}
+{"loss": 0.6107, "learning_rate": 0.0001019104377854517, "epoch": 1.32, "step": 35680}
+{"loss": 0.6868, "learning_rate": 0.0001018547398908321, "epoch": 1.32, "step": 35690}
+{"loss": 0.6235, "learning_rate": 0.00010179904199621254, "epoch": 1.32, "step": 35700}
+{"loss": 0.5671, "learning_rate": 0.00010174334410159295, "epoch": 1.32, "step": 35710}
+{"loss": 0.5926, "learning_rate": 0.00010168764620697337, "epoch": 1.32, "step": 35720}
+{"loss": 0.5502, "learning_rate": 0.00010163194831235379, "epoch": 1.32, "step": 35730}
+{"loss": 0.6109, "learning_rate": 0.0001015762504177342, "epoch": 1.32, "step": 35740}
+{"loss": 0.6107, "learning_rate": 0.00010152055252311461, "epoch": 1.33, "step": 35750}
+{"loss": 0.776, "learning_rate": 0.00010146485462849504, "epoch": 1.33, "step": 35760}
+{"loss": 0.5754, "learning_rate": 0.00010140915673387545, "epoch": 1.33, "step": 35770}
+{"loss": 0.6007, "learning_rate": 0.00010135345883925586, "epoch": 1.33, "step": 35780}
+{"loss": 0.5029, "learning_rate": 0.0001012977609446363, "epoch": 1.33, "step": 35790}
+{"loss": 0.6831, "learning_rate": 0.00010124206305001671, "epoch": 1.33, "step": 35800}
+{"loss": 0.6357, "learning_rate": 0.00010118636515539712, "epoch": 1.33, "step": 35810}
+{"loss": 0.6079, "learning_rate": 0.00010113066726077753, "epoch": 1.33, "step": 35820}
+{"loss": 0.5962, "learning_rate": 0.00010107496936615796, "epoch": 1.33, "step": 35830}
+{"loss": 0.6177, "learning_rate": 0.00010101927147153837, "epoch": 1.33, "step": 35840}
+{"loss": 0.4786, "learning_rate": 0.00010096357357691878, "epoch": 1.33, "step": 35850}
+{"loss": 0.5954, "learning_rate": 0.0001009078756822992, "epoch": 1.33, "step": 35860}
+{"loss": 0.645, "learning_rate": 0.00010085217778767961, "epoch": 1.33, "step": 35870}
+{"loss": 0.5621, "learning_rate": 0.00010079647989306003, "epoch": 1.33, "step": 35880}
+{"loss": 0.5971, "learning_rate": 0.00010074078199844045, "epoch": 1.33, "step": 35890}
+{"loss": 0.5841, "learning_rate": 0.00010068508410382086, "epoch": 1.33, "step": 35900}
+{"loss": 0.4793, "learning_rate": 0.00010062938620920129, "epoch": 1.33, "step": 35910}
+{"loss": 0.4483, "learning_rate": 0.00010057368831458171, "epoch": 1.33, "step": 35920}
+{"loss": 0.6261, "learning_rate": 0.00010051799041996212, "epoch": 1.33, "step": 35930}
+{"loss": 0.6121, "learning_rate": 0.00010046229252534253, "epoch": 1.33, "step": 35940}
+{"loss": 0.6607, "learning_rate": 0.00010040659463072294, "epoch": 1.33, "step": 35950}
+{"loss": 0.546, "learning_rate": 0.00010035089673610337, "epoch": 1.33, "step": 35960}
+{"loss": 0.6361, "learning_rate": 0.00010029519884148378, "epoch": 1.33, "step": 35970}
+{"loss": 0.6022, "learning_rate": 0.00010023950094686419, "epoch": 1.33, "step": 35980}
+{"loss": 0.5962, "learning_rate": 0.00010018380305224462, "epoch": 1.33, "step": 35990}
+{"loss": 0.6776, "learning_rate": 0.00010012810515762503, "epoch": 1.33, "step": 36000}
+{"loss": 0.4473, "learning_rate": 0.00010007240726300544, "epoch": 1.33, "step": 36010}
+{"loss": 0.5978, "learning_rate": 0.00010001670936838588, "epoch": 1.34, "step": 36020}
+{"loss": 0.5636, "learning_rate": 9.996101147376629e-05, "epoch": 1.34, "step": 36030}
+{"loss": 0.5971, "learning_rate": 9.99053135791467e-05, "epoch": 1.34, "step": 36040}
+{"loss": 0.6218, "learning_rate": 9.984961568452712e-05, "epoch": 1.34, "step": 36050}
+{"loss": 0.4685, "learning_rate": 9.979391778990753e-05, "epoch": 1.34, "step": 36060}
+{"loss": 0.5591, "learning_rate": 9.973821989528795e-05, "epoch": 1.34, "step": 36070}
+{"loss": 0.4999, "learning_rate": 9.968252200066836e-05, "epoch": 1.34, "step": 36080}
+{"loss": 0.5702, "learning_rate": 9.962682410604878e-05, "epoch": 1.34, "step": 36090}
+{"loss": 0.503, "learning_rate": 9.957112621142919e-05, "epoch": 1.34, "step": 36100}
+{"loss": 0.5753, "learning_rate": 9.95154283168096e-05, "epoch": 1.34, "step": 36110}
+{"loss": 0.5989, "learning_rate": 9.945973042219004e-05, "epoch": 1.34, "step": 36120}
+{"loss": 0.5798, "learning_rate": 9.940403252757045e-05, "epoch": 1.34, "step": 36130}
+{"loss": 0.5491, "learning_rate": 9.934833463295086e-05, "epoch": 1.34, "step": 36140}
+{"loss": 0.4819, "learning_rate": 9.929263673833129e-05, "epoch": 1.34, "step": 36150}
+{"loss": 0.6227, "learning_rate": 9.92369388437117e-05, "epoch": 1.34, "step": 36160}
+{"loss": 0.6807, "learning_rate": 9.918124094909211e-05, "epoch": 1.34, "step": 36170}
+{"loss": 0.5521, "learning_rate": 9.912554305447254e-05, "epoch": 1.34, "step": 36180}
+{"loss": 0.5717, "learning_rate": 9.906984515985295e-05, "epoch": 1.34, "step": 36190}
+{"loss": 0.5666, "learning_rate": 9.901414726523336e-05, "epoch": 1.34, "step": 36200}
+{"loss": 0.5859, "learning_rate": 9.89584493706138e-05, "epoch": 1.34, "step": 36210}
+{"loss": 0.6087, "learning_rate": 9.890275147599421e-05, "epoch": 1.34, "step": 36220}
+{"loss": 0.6007, "learning_rate": 9.884705358137462e-05, "epoch": 1.34, "step": 36230}
+{"loss": 0.5538, "learning_rate": 9.879135568675503e-05, "epoch": 1.34, "step": 36240}
+{"loss": 0.5587, "learning_rate": 9.873565779213545e-05, "epoch": 1.34, "step": 36250}
+{"loss": 0.5805, "learning_rate": 9.867995989751587e-05, "epoch": 1.34, "step": 36260}
+{"loss": 0.5035, "learning_rate": 9.862426200289628e-05, "epoch": 1.34, "step": 36270}
+{"loss": 0.5851, "learning_rate": 9.85685641082767e-05, "epoch": 1.34, "step": 36280}
+{"loss": 0.6262, "learning_rate": 9.851286621365711e-05, "epoch": 1.35, "step": 36290}
+{"loss": 0.4777, "learning_rate": 9.845716831903752e-05, "epoch": 1.35, "step": 36300}
+{"loss": 0.4962, "learning_rate": 9.840147042441796e-05, "epoch": 1.35, "step": 36310}
+{"loss": 0.4893, "learning_rate": 9.834577252979837e-05, "epoch": 1.35, "step": 36320}
+{"loss": 0.541, "learning_rate": 9.829007463517878e-05, "epoch": 1.35, "step": 36330}
+{"loss": 0.6842, "learning_rate": 9.823437674055921e-05, "epoch": 1.35, "step": 36340}
+{"loss": 0.6965, "learning_rate": 9.817867884593962e-05, "epoch": 1.35, "step": 36350}
+{"loss": 0.4909, "learning_rate": 9.812298095132003e-05, "epoch": 1.35, "step": 36360}
+{"loss": 0.5554, "learning_rate": 9.806728305670044e-05, "epoch": 1.35, "step": 36370}
+{"loss": 0.589, "learning_rate": 9.801158516208087e-05, "epoch": 1.35, "step": 36380}
+{"loss": 0.6397, "learning_rate": 9.795588726746128e-05, "epoch": 1.35, "step": 36390}
+{"loss": 0.5537, "learning_rate": 9.790018937284169e-05, "epoch": 1.35, "step": 36400}
+{"loss": 0.6147, "learning_rate": 9.784449147822211e-05, "epoch": 1.35, "step": 36410}
+{"loss": 0.7166, "learning_rate": 9.778879358360254e-05, "epoch": 1.35, "step": 36420}
+{"loss": 0.592, "learning_rate": 9.773309568898295e-05, "epoch": 1.35, "step": 36430}
+{"loss": 0.5838, "learning_rate": 9.767739779436337e-05, "epoch": 1.35, "step": 36440}
+{"loss": 0.571, "learning_rate": 9.762169989974379e-05, "epoch": 1.35, "step": 36450}
+{"loss": 0.7476, "learning_rate": 9.75660020051242e-05, "epoch": 1.35, "step": 36460}
+{"loss": 0.4528, "learning_rate": 9.751030411050462e-05, "epoch": 1.35, "step": 36470}
+{"loss": 0.554, "learning_rate": 9.745460621588503e-05, "epoch": 1.35, "step": 36480}
+{"loss": 0.5297, "learning_rate": 9.739890832126544e-05, "epoch": 1.35, "step": 36490}
+{"loss": 0.6112, "learning_rate": 9.734321042664585e-05, "epoch": 1.35, "step": 36500}
+{"loss": 0.4328, "learning_rate": 9.728751253202628e-05, "epoch": 1.35, "step": 36510}
+{"loss": 0.5773, "learning_rate": 9.723181463740669e-05, "epoch": 1.35, "step": 36520}
+{"loss": 0.6562, "learning_rate": 9.71761167427871e-05, "epoch": 1.35, "step": 36530}
+{"loss": 0.5885, "learning_rate": 9.712041884816754e-05, "epoch": 1.35, "step": 36540}
+{"loss": 0.5799, "learning_rate": 9.706472095354795e-05, "epoch": 1.35, "step": 36550}
+{"loss": 0.5385, "learning_rate": 9.700902305892836e-05, "epoch": 1.36, "step": 36560}
+{"loss": 0.7173, "learning_rate": 9.695332516430879e-05, "epoch": 1.36, "step": 36570}
+{"loss": 0.4672, "learning_rate": 9.68976272696892e-05, "epoch": 1.36, "step": 36580}
+{"loss": 0.666, "learning_rate": 9.684192937506961e-05, "epoch": 1.36, "step": 36590}
+{"loss": 0.5006, "learning_rate": 9.678623148045003e-05, "epoch": 1.36, "step": 36600}
+{"loss": 0.5442, "learning_rate": 9.673053358583044e-05, "epoch": 1.36, "step": 36610}
+{"loss": 0.5064, "learning_rate": 9.667483569121086e-05, "epoch": 1.36, "step": 36620}
+{"loss": 0.5305, "learning_rate": 9.662470758605323e-05, "epoch": 1.36, "step": 36630}
+{"loss": 0.575, "learning_rate": 9.656900969143367e-05, "epoch": 1.36, "step": 36640}
+{"loss": 0.4911, "learning_rate": 9.651331179681408e-05, "epoch": 1.36, "step": 36650}
+{"loss": 0.5857, "learning_rate": 9.645761390219449e-05, "epoch": 1.36, "step": 36660}
+{"loss": 0.5431, "learning_rate": 9.64019160075749e-05, "epoch": 1.36, "step": 36670}
+{"loss": 0.5609, "learning_rate": 9.634621811295532e-05, "epoch": 1.36, "step": 36680}
+{"loss": 0.6122, "learning_rate": 9.629052021833574e-05, "epoch": 1.36, "step": 36690}
+{"loss": 0.6579, "learning_rate": 9.623482232371615e-05, "epoch": 1.36, "step": 36700}
+{"loss": 0.6233, "learning_rate": 9.617912442909657e-05, "epoch": 1.36, "step": 36710}
+{"loss": 0.6777, "learning_rate": 9.612342653447698e-05, "epoch": 1.36, "step": 36720}
+{"loss": 0.5071, "learning_rate": 9.60677286398574e-05, "epoch": 1.36, "step": 36730}
+{"loss": 0.6418, "learning_rate": 9.601203074523783e-05, "epoch": 1.36, "step": 36740}
+{"loss": 0.5162, "learning_rate": 9.595633285061824e-05, "epoch": 1.36, "step": 36750}
+{"loss": 0.6013, "learning_rate": 9.590063495599865e-05, "epoch": 1.36, "step": 36760}
+{"loss": 0.487, "learning_rate": 9.584493706137908e-05, "epoch": 1.36, "step": 36770}
+{"loss": 0.5323, "learning_rate": 9.578923916675949e-05, "epoch": 1.36, "step": 36780}
+{"loss": 0.5625, "learning_rate": 9.57335412721399e-05, "epoch": 1.36, "step": 36790}
+{"loss": 0.5449, "learning_rate": 9.567784337752031e-05, "epoch": 1.36, "step": 36800}
+{"loss": 0.653, "learning_rate": 9.562214548290074e-05, "epoch": 1.36, "step": 36810}
+{"loss": 0.5442, "learning_rate": 9.556644758828115e-05, "epoch": 1.36, "step": 36820}
+{"loss": 0.745, "learning_rate": 9.551074969366156e-05, "epoch": 1.37, "step": 36830}
+{"loss": 0.5711, "learning_rate": 9.5455051799042e-05, "epoch": 1.37, "step": 36840}
+{"loss": 0.643, "learning_rate": 9.539935390442241e-05, "epoch": 1.37, "step": 36850}
+{"loss": 0.6256, "learning_rate": 9.534365600980282e-05, "epoch": 1.37, "step": 36860}
+{"loss": 0.6328, "learning_rate": 9.528795811518324e-05, "epoch": 1.37, "step": 36870}
+{"loss": 0.6809, "learning_rate": 9.523226022056366e-05, "epoch": 1.37, "step": 36880}
+{"loss": 0.5442, "learning_rate": 9.517656232594407e-05, "epoch": 1.37, "step": 36890}
+{"loss": 0.5729, "learning_rate": 9.512086443132449e-05, "epoch": 1.37, "step": 36900}
+{"loss": 0.5166, "learning_rate": 9.50651665367049e-05, "epoch": 1.37, "step": 36910}
+{"loss": 0.4517, "learning_rate": 9.500946864208531e-05, "epoch": 1.37, "step": 36920}
+{"loss": 0.6478, "learning_rate": 9.495377074746573e-05, "epoch": 1.37, "step": 36930}
+{"loss": 0.5478, "learning_rate": 9.489807285284616e-05, "epoch": 1.37, "step": 36940}
+{"loss": 0.5259, "learning_rate": 9.484237495822657e-05, "epoch": 1.37, "step": 36950}
+{"loss": 0.5417, "learning_rate": 9.478667706360699e-05, "epoch": 1.37, "step": 36960}
+{"loss": 0.5827, "learning_rate": 9.473097916898741e-05, "epoch": 1.37, "step": 36970}
+{"loss": 0.609, "learning_rate": 9.467528127436782e-05, "epoch": 1.37, "step": 36980}
+{"loss": 0.6195, "learning_rate": 9.461958337974823e-05, "epoch": 1.37, "step": 36990}
+{"loss": 0.6875, "learning_rate": 9.456388548512866e-05, "epoch": 1.37, "step": 37000}
+{"loss": 0.6355, "learning_rate": 9.450818759050907e-05, "epoch": 1.37, "step": 37010}
+{"loss": 0.547, "learning_rate": 9.445248969588948e-05, "epoch": 1.37, "step": 37020}
+{"loss": 0.4833, "learning_rate": 9.43967918012699e-05, "epoch": 1.37, "step": 37030}
+{"loss": 0.7239, "learning_rate": 9.434109390665033e-05, "epoch": 1.37, "step": 37040}
+{"loss": 0.5379, "learning_rate": 9.428539601203074e-05, "epoch": 1.37, "step": 37050}
+{"loss": 0.6801, "learning_rate": 9.422969811741115e-05, "epoch": 1.37, "step": 37060}
+{"loss": 0.4889, "learning_rate": 9.417400022279158e-05, "epoch": 1.37, "step": 37070}
+{"loss": 0.6934, "learning_rate": 9.411830232817199e-05, "epoch": 1.37, "step": 37080}
+{"loss": 0.6061, "learning_rate": 9.40626044335524e-05, "epoch": 1.37, "step": 37090}
+{"loss": 0.4343, "learning_rate": 9.400690653893282e-05, "epoch": 1.38, "step": 37100}
+{"loss": 0.6552, "learning_rate": 9.395120864431323e-05, "epoch": 1.38, "step": 37110}
+{"loss": 0.6288, "learning_rate": 9.389551074969365e-05, "epoch": 1.38, "step": 37120}
+{"loss": 0.4945, "learning_rate": 9.383981285507407e-05, "epoch": 1.38, "step": 37130}
+{"loss": 0.492, "learning_rate": 9.378411496045448e-05, "epoch": 1.38, "step": 37140}
+{"loss": 0.5001, "learning_rate": 9.372841706583489e-05, "epoch": 1.38, "step": 37150}
+{"loss": 0.5464, "learning_rate": 9.367271917121533e-05, "epoch": 1.38, "step": 37160}
+{"loss": 0.5427, "learning_rate": 9.361702127659574e-05, "epoch": 1.38, "step": 37170}
+{"loss": 0.5644, "learning_rate": 9.356132338197615e-05, "epoch": 1.38, "step": 37180}
+{"loss": 0.5245, "learning_rate": 9.350562548735658e-05, "epoch": 1.38, "step": 37190}
+{"loss": 0.6931, "learning_rate": 9.344992759273699e-05, "epoch": 1.38, "step": 37200}
+{"loss": 0.5938, "learning_rate": 9.33942296981174e-05, "epoch": 1.38, "step": 37210}
+{"loss": 0.502, "learning_rate": 9.333853180349781e-05, "epoch": 1.38, "step": 37220}
+{"loss": 0.5595, "learning_rate": 9.328283390887824e-05, "epoch": 1.38, "step": 37230}
+{"loss": 0.6337, "learning_rate": 9.322713601425865e-05, "epoch": 1.38, "step": 37240}
+{"loss": 0.5674, "learning_rate": 9.317143811963906e-05, "epoch": 1.38, "step": 37250}
+{"loss": 0.6151, "learning_rate": 9.31157402250195e-05, "epoch": 1.38, "step": 37260}
+{"loss": 0.6653, "learning_rate": 9.306004233039991e-05, "epoch": 1.38, "step": 37270}
+{"loss": 0.566, "learning_rate": 9.300434443578032e-05, "epoch": 1.38, "step": 37280}
+{"loss": 0.629, "learning_rate": 9.294864654116074e-05, "epoch": 1.38, "step": 37290}
+{"loss": 0.6316, "learning_rate": 9.289294864654115e-05, "epoch": 1.38, "step": 37300}
+{"loss": 0.6943, "learning_rate": 9.284282054138353e-05, "epoch": 1.38, "step": 37310}
+{"loss": 0.5991, "learning_rate": 9.278712264676394e-05, "epoch": 1.38, "step": 37320}
+{"loss": 0.5962, "learning_rate": 9.273142475214436e-05, "epoch": 1.38, "step": 37330}
+{"loss": 0.6278, "learning_rate": 9.267572685752477e-05, "epoch": 1.38, "step": 37340}
+{"loss": 0.584, "learning_rate": 9.262002896290519e-05, "epoch": 1.38, "step": 37350}
+{"loss": 0.5016, "learning_rate": 9.256433106828562e-05, "epoch": 1.38, "step": 37360}
+{"loss": 0.7003, "learning_rate": 9.250863317366603e-05, "epoch": 1.39, "step": 37370}
+{"loss": 0.4483, "learning_rate": 9.245293527904645e-05, "epoch": 1.39, "step": 37380}
+{"loss": 0.52, "learning_rate": 9.239723738442686e-05, "epoch": 1.39, "step": 37390}
+{"loss": 0.588, "learning_rate": 9.234153948980728e-05, "epoch": 1.39, "step": 37400}
+{"loss": 0.5803, "learning_rate": 9.228584159518769e-05, "epoch": 1.39, "step": 37410}
+{"loss": 0.5044, "learning_rate": 9.22301437005681e-05, "epoch": 1.39, "step": 37420}
+{"loss": 0.4983, "learning_rate": 9.217444580594853e-05, "epoch": 1.39, "step": 37430}
+{"loss": 0.5265, "learning_rate": 9.211874791132894e-05, "epoch": 1.39, "step": 37440}
+{"loss": 0.5209, "learning_rate": 9.206305001670935e-05, "epoch": 1.39, "step": 37450}
+{"loss": 0.4863, "learning_rate": 9.200735212208979e-05, "epoch": 1.39, "step": 37460}
+{"loss": 0.7458, "learning_rate": 9.19516542274702e-05, "epoch": 1.39, "step": 37470}
+{"loss": 0.6365, "learning_rate": 9.189595633285061e-05, "epoch": 1.39, "step": 37480}
+{"loss": 0.6431, "learning_rate": 9.184025843823104e-05, "epoch": 1.39, "step": 37490}
+{"loss": 0.6689, "learning_rate": 9.178456054361145e-05, "epoch": 1.39, "step": 37500}
+{"loss": 0.5844, "learning_rate": 9.172886264899186e-05, "epoch": 1.39, "step": 37510}
+{"loss": 0.6368, "learning_rate": 9.167316475437227e-05, "epoch": 1.39, "step": 37520}
+{"loss": 0.6164, "learning_rate": 9.16174668597527e-05, "epoch": 1.39, "step": 37530}
+{"loss": 0.6848, "learning_rate": 9.15617689651331e-05, "epoch": 1.39, "step": 37540}
+{"loss": 0.6895, "learning_rate": 9.150607107051352e-05, "epoch": 1.39, "step": 37550}
+{"loss": 0.6693, "learning_rate": 9.145037317589395e-05, "epoch": 1.39, "step": 37560}
+{"loss": 0.5838, "learning_rate": 9.139467528127437e-05, "epoch": 1.39, "step": 37570}
+{"loss": 0.6268, "learning_rate": 9.133897738665478e-05, "epoch": 1.39, "step": 37580}
+{"loss": 0.5345, "learning_rate": 9.12832794920352e-05, "epoch": 1.39, "step": 37590}
+{"loss": 0.6347, "learning_rate": 9.122758159741561e-05, "epoch": 1.39, "step": 37600}
+{"loss": 0.6112, "learning_rate": 9.117188370279602e-05, "epoch": 1.39, "step": 37610}
+{"loss": 0.5432, "learning_rate": 9.111618580817645e-05, "epoch": 1.39, "step": 37620}
+{"loss": 0.5615, "learning_rate": 9.106048791355686e-05, "epoch": 1.39, "step": 37630}
+{"loss": 0.6325, "learning_rate": 9.100479001893727e-05, "epoch": 1.4, "step": 37640}
+{"loss": 0.6287, "learning_rate": 9.094909212431768e-05, "epoch": 1.4, "step": 37650}
+{"loss": 0.6164, "learning_rate": 9.089339422969812e-05, "epoch": 1.4, "step": 37660}
+{"loss": 0.5449, "learning_rate": 9.083769633507853e-05, "epoch": 1.4, "step": 37670}
+{"loss": 0.5394, "learning_rate": 9.078199844045894e-05, "epoch": 1.4, "step": 37680}
+{"loss": 0.5274, "learning_rate": 9.072630054583937e-05, "epoch": 1.4, "step": 37690}
+{"loss": 0.5024, "learning_rate": 9.067060265121978e-05, "epoch": 1.4, "step": 37700}
+{"loss": 0.5863, "learning_rate": 9.061490475660019e-05, "epoch": 1.4, "step": 37710}
+{"loss": 0.5523, "learning_rate": 9.055920686198061e-05, "epoch": 1.4, "step": 37720}
+{"loss": 0.6241, "learning_rate": 9.050350896736103e-05, "epoch": 1.4, "step": 37730}
+{"loss": 0.5805, "learning_rate": 9.044781107274144e-05, "epoch": 1.4, "step": 37740}
+{"loss": 0.6348, "learning_rate": 9.039211317812186e-05, "epoch": 1.4, "step": 37750}
+{"loss": 0.5918, "learning_rate": 9.033641528350227e-05, "epoch": 1.4, "step": 37760}
+{"loss": 0.6979, "learning_rate": 9.028071738888268e-05, "epoch": 1.4, "step": 37770}
+{"loss": 0.4575, "learning_rate": 9.022501949426311e-05, "epoch": 1.4, "step": 37780}
+{"loss": 0.4958, "learning_rate": 9.016932159964353e-05, "epoch": 1.4, "step": 37790}
+{"loss": 0.5361, "learning_rate": 9.011362370502394e-05, "epoch": 1.4, "step": 37800}
+{"loss": 0.599, "learning_rate": 9.005792581040435e-05, "epoch": 1.4, "step": 37810}
+{"loss": 0.589, "learning_rate": 9.000222791578478e-05, "epoch": 1.4, "step": 37820}
+{"loss": 0.7167, "learning_rate": 8.994653002116519e-05, "epoch": 1.4, "step": 37830}
+{"loss": 0.6201, "learning_rate": 8.98908321265456e-05, "epoch": 1.4, "step": 37840}
+{"loss": 0.4408, "learning_rate": 8.983513423192603e-05, "epoch": 1.4, "step": 37850}
+{"loss": 0.624, "learning_rate": 8.977943633730644e-05, "epoch": 1.4, "step": 37860}
+{"loss": 0.5171, "learning_rate": 8.972373844268685e-05, "epoch": 1.4, "step": 37870}
+{"loss": 0.6237, "learning_rate": 8.966804054806729e-05, "epoch": 1.4, "step": 37880}
+{"loss": 0.4729, "learning_rate": 8.96123426534477e-05, "epoch": 1.4, "step": 37890}
+{"loss": 0.6422, "learning_rate": 8.955664475882811e-05, "epoch": 1.4, "step": 37900}
+{"loss": 0.5589, "learning_rate": 8.950094686420852e-05, "epoch": 1.41, "step": 37910}
+{"loss": 0.5995, "learning_rate": 8.944524896958895e-05, "epoch": 1.41, "step": 37920}
+{"loss": 0.4762, "learning_rate": 8.938955107496936e-05, "epoch": 1.41, "step": 37930}
+{"loss": 0.5674, "learning_rate": 8.933385318034977e-05, "epoch": 1.41, "step": 37940}
+{"loss": 0.6598, "learning_rate": 8.927815528573019e-05, "epoch": 1.41, "step": 37950}
+{"loss": 0.512, "learning_rate": 8.92224573911106e-05, "epoch": 1.41, "step": 37960}
+{"loss": 0.5665, "learning_rate": 8.916675949649101e-05, "epoch": 1.41, "step": 37970}
+{"loss": 0.472, "learning_rate": 8.911106160187145e-05, "epoch": 1.41, "step": 37980}
+{"loss": 0.5788, "learning_rate": 8.905536370725186e-05, "epoch": 1.41, "step": 37990}
+{"loss": 0.5859, "learning_rate": 8.899966581263227e-05, "epoch": 1.41, "step": 38000}
+{"loss": 0.5586, "learning_rate": 8.89439679180127e-05, "epoch": 1.41, "step": 38010}
+{"loss": 0.4824, "learning_rate": 8.888827002339311e-05, "epoch": 1.41, "step": 38020}
+{"loss": 0.5916, "learning_rate": 8.883257212877352e-05, "epoch": 1.41, "step": 38030}
+{"loss": 0.6081, "learning_rate": 8.877687423415393e-05, "epoch": 1.41, "step": 38040}
+{"loss": 0.6108, "learning_rate": 8.872117633953436e-05, "epoch": 1.41, "step": 38050}
+{"loss": 0.6746, "learning_rate": 8.866547844491477e-05, "epoch": 1.41, "step": 38060}
+{"loss": 0.6062, "learning_rate": 8.860978055029518e-05, "epoch": 1.41, "step": 38070}
+{"loss": 0.658, "learning_rate": 8.855408265567562e-05, "epoch": 1.41, "step": 38080}
+{"loss": 0.5576, "learning_rate": 8.849838476105603e-05, "epoch": 1.41, "step": 38090}
+{"loss": 0.6388, "learning_rate": 8.844268686643644e-05, "epoch": 1.41, "step": 38100}
+{"loss": 0.5272, "learning_rate": 8.838698897181687e-05, "epoch": 1.41, "step": 38110}
+{"loss": 0.5783, "learning_rate": 8.833129107719728e-05, "epoch": 1.41, "step": 38120}
+{"loss": 0.6013, "learning_rate": 8.827559318257769e-05, "epoch": 1.41, "step": 38130}
+{"loss": 0.6183, "learning_rate": 8.821989528795811e-05, "epoch": 1.41, "step": 38140}
+{"loss": 0.5121, "learning_rate": 8.816419739333852e-05, "epoch": 1.41, "step": 38150}
+{"loss": 0.7646, "learning_rate": 8.810849949871893e-05, "epoch": 1.41, "step": 38160}
+{"loss": 0.7093, "learning_rate": 8.805280160409936e-05, "epoch": 1.41, "step": 38170}
+{"loss": 0.7307, "learning_rate": 8.799710370947978e-05, "epoch": 1.42, "step": 38180}
+{"loss": 0.578, "learning_rate": 8.79414058148602e-05, "epoch": 1.42, "step": 38190}
+{"loss": 0.6395, "learning_rate": 8.78857079202406e-05, "epoch": 1.42, "step": 38200}
+{"loss": 0.4526, "learning_rate": 8.783001002562103e-05, "epoch": 1.42, "step": 38210}
+{"loss": 0.7063, "learning_rate": 8.777431213100144e-05, "epoch": 1.42, "step": 38220}
+{"loss": 0.4838, "learning_rate": 8.771861423638185e-05, "epoch": 1.42, "step": 38230}
+{"loss": 0.5524, "learning_rate": 8.766291634176228e-05, "epoch": 1.42, "step": 38240}
+{"loss": 0.4842, "learning_rate": 8.760721844714269e-05, "epoch": 1.42, "step": 38250}
+{"loss": 0.5478, "learning_rate": 8.75515205525231e-05, "epoch": 1.42, "step": 38260}
+{"loss": 0.6879, "learning_rate": 8.749582265790352e-05, "epoch": 1.42, "step": 38270}
+{"loss": 0.6129, "learning_rate": 8.744012476328394e-05, "epoch": 1.42, "step": 38280}
+{"loss": 0.7345, "learning_rate": 8.738442686866435e-05, "epoch": 1.42, "step": 38290}
+{"loss": 0.6483, "learning_rate": 8.732872897404479e-05, "epoch": 1.42, "step": 38300}
+{"loss": 0.5696, "learning_rate": 8.72730310794252e-05, "epoch": 1.42, "step": 38310}
+{"loss": 0.562, "learning_rate": 8.721733318480561e-05, "epoch": 1.42, "step": 38320}
+{"loss": 0.5372, "learning_rate": 8.716163529018602e-05, "epoch": 1.42, "step": 38330}
+{"loss": 0.5587, "learning_rate": 8.710593739556644e-05, "epoch": 1.42, "step": 38340}
+{"loss": 0.6179, "learning_rate": 8.705023950094685e-05, "epoch": 1.42, "step": 38350}
+{"loss": 0.5569, "learning_rate": 8.699454160632727e-05, "epoch": 1.42, "step": 38360}
+{"loss": 0.5788, "learning_rate": 8.693884371170769e-05, "epoch": 1.42, "step": 38370}
+{"loss": 0.573, "learning_rate": 8.68831458170881e-05, "epoch": 1.42, "step": 38380}
+{"loss": 0.6139, "learning_rate": 8.682744792246851e-05, "epoch": 1.42, "step": 38390}
+{"loss": 0.6121, "learning_rate": 8.677175002784895e-05, "epoch": 1.42, "step": 38400}
+{"loss": 0.5259, "learning_rate": 8.671605213322936e-05, "epoch": 1.42, "step": 38410}
+{"loss": 0.588, "learning_rate": 8.666035423860977e-05, "epoch": 1.42, "step": 38420}
+{"loss": 0.4979, "learning_rate": 8.66046563439902e-05, "epoch": 1.42, "step": 38430}
+{"loss": 0.6504, "learning_rate": 8.654895844937061e-05, "epoch": 1.42, "step": 38440}
+{"loss": 0.6656, "learning_rate": 8.649326055475102e-05, "epoch": 1.43, "step": 38450}
+{"loss": 0.5916, "learning_rate": 8.643756266013143e-05, "epoch": 1.43, "step": 38460}
+{"loss": 0.5339, "learning_rate": 8.638186476551186e-05, "epoch": 1.43, "step": 38470}
+{"loss": 0.6233, "learning_rate": 8.632616687089227e-05, "epoch": 1.43, "step": 38480}
+{"loss": 0.5343, "learning_rate": 8.627046897627268e-05, "epoch": 1.43, "step": 38490}
+{"loss": 0.5898, "learning_rate": 8.621477108165312e-05, "epoch": 1.43, "step": 38500}
+{"loss": 0.4948, "learning_rate": 8.615907318703353e-05, "epoch": 1.43, "step": 38510}
+{"loss": 0.6933, "learning_rate": 8.610337529241394e-05, "epoch": 1.43, "step": 38520}
+{"loss": 0.5161, "learning_rate": 8.604767739779436e-05, "epoch": 1.43, "step": 38530}
+{"loss": 0.6217, "learning_rate": 8.599197950317477e-05, "epoch": 1.43, "step": 38540}
+{"loss": 0.5585, "learning_rate": 8.593628160855519e-05, "epoch": 1.43, "step": 38550}
+{"loss": 0.7432, "learning_rate": 8.588058371393561e-05, "epoch": 1.43, "step": 38560}
+{"loss": 0.7917, "learning_rate": 8.582488581931602e-05, "epoch": 1.43, "step": 38570}
+{"loss": 0.6663, "learning_rate": 8.576918792469643e-05, "epoch": 1.43, "step": 38580}
+{"loss": 0.6414, "learning_rate": 8.571349003007684e-05, "epoch": 1.43, "step": 38590}
+{"loss": 0.6547, "learning_rate": 8.565779213545728e-05, "epoch": 1.43, "step": 38600}
+{"loss": 0.4906, "learning_rate": 8.560209424083769e-05, "epoch": 1.43, "step": 38610}
+{"loss": 0.6133, "learning_rate": 8.55463963462181e-05, "epoch": 1.43, "step": 38620}
+{"loss": 0.6114, "learning_rate": 8.549069845159853e-05, "epoch": 1.43, "step": 38630}
+{"loss": 0.6121, "learning_rate": 8.543500055697894e-05, "epoch": 1.43, "step": 38640}
+{"loss": 0.5713, "learning_rate": 8.537930266235935e-05, "epoch": 1.43, "step": 38650}
+{"loss": 0.6192, "learning_rate": 8.532360476773978e-05, "epoch": 1.43, "step": 38660}
+{"loss": 0.5984, "learning_rate": 8.526790687312019e-05, "epoch": 1.43, "step": 38670}
+{"loss": 0.5022, "learning_rate": 8.52122089785006e-05, "epoch": 1.43, "step": 38680}
+{"loss": 0.594, "learning_rate": 8.515651108388102e-05, "epoch": 1.43, "step": 38690}
+{"loss": 0.6233, "learning_rate": 8.510081318926145e-05, "epoch": 1.43, "step": 38700}
+{"loss": 0.49, "learning_rate": 8.504511529464186e-05, "epoch": 1.43, "step": 38710}
+{"loss": 0.5534, "learning_rate": 8.498941740002227e-05, "epoch": 1.44, "step": 38720}
+{"loss": 0.5978, "learning_rate": 8.49337195054027e-05, "epoch": 1.44, "step": 38730}
+{"loss": 0.5747, "learning_rate": 8.48780216107831e-05, "epoch": 1.44, "step": 38740}
+{"loss": 0.6752, "learning_rate": 8.482232371616352e-05, "epoch": 1.44, "step": 38750}
+{"loss": 0.6899, "learning_rate": 8.476662582154394e-05, "epoch": 1.44, "step": 38760}
+{"loss": 0.4496, "learning_rate": 8.471092792692435e-05, "epoch": 1.44, "step": 38770}
+{"loss": 0.4616, "learning_rate": 8.465523003230476e-05, "epoch": 1.44, "step": 38780}
+{"loss": 0.6139, "learning_rate": 8.459953213768519e-05, "epoch": 1.44, "step": 38790}
+{"loss": 0.4745, "learning_rate": 8.45438342430656e-05, "epoch": 1.44, "step": 38800}
+{"loss": 0.6267, "learning_rate": 8.448813634844601e-05, "epoch": 1.44, "step": 38810}
+{"loss": 0.4762, "learning_rate": 8.443243845382645e-05, "epoch": 1.44, "step": 38820}
+{"loss": 0.6528, "learning_rate": 8.437674055920686e-05, "epoch": 1.44, "step": 38830}
+{"loss": 0.6377, "learning_rate": 8.432104266458727e-05, "epoch": 1.44, "step": 38840}
+{"loss": 0.5159, "learning_rate": 8.426534476996768e-05, "epoch": 1.44, "step": 38850}
+{"loss": 0.4913, "learning_rate": 8.420964687534811e-05, "epoch": 1.44, "step": 38860}
+{"loss": 0.7115, "learning_rate": 8.415394898072852e-05, "epoch": 1.44, "step": 38870}
+{"loss": 0.4628, "learning_rate": 8.409825108610893e-05, "epoch": 1.44, "step": 38880}
+{"loss": 0.6262, "learning_rate": 8.404255319148935e-05, "epoch": 1.44, "step": 38890}
+{"loss": 0.6766, "learning_rate": 8.398685529686976e-05, "epoch": 1.44, "step": 38900}
+{"loss": 0.4706, "learning_rate": 8.393115740225018e-05, "epoch": 1.44, "step": 38910}
+{"loss": 0.646, "learning_rate": 8.387545950763061e-05, "epoch": 1.44, "step": 38920}
+{"loss": 0.5226, "learning_rate": 8.381976161301103e-05, "epoch": 1.44, "step": 38930}
+{"loss": 0.4093, "learning_rate": 8.376406371839144e-05, "epoch": 1.44, "step": 38940}
+{"loss": 0.5973, "learning_rate": 8.370836582377186e-05, "epoch": 1.44, "step": 38950}
+{"loss": 0.7618, "learning_rate": 8.365266792915227e-05, "epoch": 1.44, "step": 38960}
+{"loss": 0.6285, "learning_rate": 8.359697003453268e-05, "epoch": 1.44, "step": 38970}
+{"loss": 0.5303, "learning_rate": 8.35412721399131e-05, "epoch": 1.44, "step": 38980}
+{"loss": 0.6095, "learning_rate": 8.348557424529352e-05, "epoch": 1.45, "step": 38990}
+{"loss": 0.5461, "learning_rate": 8.342987635067393e-05, "epoch": 1.45, "step": 39000}
+{"loss": 0.5979, "learning_rate": 8.337417845605434e-05, "epoch": 1.45, "step": 39010}
+{"loss": 0.6598, "learning_rate": 8.331848056143478e-05, "epoch": 1.45, "step": 39020}
+{"loss": 0.5353, "learning_rate": 8.326278266681519e-05, "epoch": 1.45, "step": 39030}
+{"loss": 0.7684, "learning_rate": 8.32070847721956e-05, "epoch": 1.45, "step": 39040}
+{"loss": 0.5507, "learning_rate": 8.315138687757603e-05, "epoch": 1.45, "step": 39050}
+{"loss": 0.5111, "learning_rate": 8.309568898295644e-05, "epoch": 1.45, "step": 39060}
+{"loss": 0.4786, "learning_rate": 8.303999108833685e-05, "epoch": 1.45, "step": 39070}
+{"loss": 0.5672, "learning_rate": 8.298429319371727e-05, "epoch": 1.45, "step": 39080}
+{"loss": 0.7179, "learning_rate": 8.292859529909768e-05, "epoch": 1.45, "step": 39090}
+{"loss": 0.5183, "learning_rate": 8.28728974044781e-05, "epoch": 1.45, "step": 39100}
+{"loss": 0.4978, "learning_rate": 8.281719950985853e-05, "epoch": 1.45, "step": 39110}
+{"loss": 0.5553, "learning_rate": 8.276150161523895e-05, "epoch": 1.45, "step": 39120}
+{"loss": 0.5143, "learning_rate": 8.270580372061936e-05, "epoch": 1.45, "step": 39130}
+{"loss": 0.5938, "learning_rate": 8.265010582599977e-05, "epoch": 1.45, "step": 39140}
+{"loss": 0.579, "learning_rate": 8.259440793138019e-05, "epoch": 1.45, "step": 39150}
+{"loss": 0.6513, "learning_rate": 8.25387100367606e-05, "epoch": 1.45, "step": 39160}
+{"loss": 0.527, "learning_rate": 8.248301214214101e-05, "epoch": 1.45, "step": 39170}
+{"loss": 0.5359, "learning_rate": 8.242731424752144e-05, "epoch": 1.45, "step": 39180}
+{"loss": 0.5852, "learning_rate": 8.237161635290185e-05, "epoch": 1.45, "step": 39190}
+{"loss": 0.6273, "learning_rate": 8.231591845828226e-05, "epoch": 1.45, "step": 39200}
+{"loss": 0.5162, "learning_rate": 8.22602205636627e-05, "epoch": 1.45, "step": 39210}
+{"loss": 0.5746, "learning_rate": 8.220452266904311e-05, "epoch": 1.45, "step": 39220}
+{"loss": 0.4622, "learning_rate": 8.214882477442352e-05, "epoch": 1.45, "step": 39230}
+{"loss": 0.5128, "learning_rate": 8.209312687980395e-05, "epoch": 1.45, "step": 39240}
+{"loss": 0.5382, "learning_rate": 8.203742898518436e-05, "epoch": 1.45, "step": 39250}
+{"loss": 0.6015, "learning_rate": 8.198173109056477e-05, "epoch": 1.46, "step": 39260}
+{"loss": 0.7172, "learning_rate": 8.192603319594518e-05, "epoch": 1.46, "step": 39270}
+{"loss": 0.6598, "learning_rate": 8.18703353013256e-05, "epoch": 1.46, "step": 39280}
+{"loss": 0.6191, "learning_rate": 8.181463740670602e-05, "epoch": 1.46, "step": 39290}
+{"loss": 0.5241, "learning_rate": 8.175893951208643e-05, "epoch": 1.46, "step": 39300}
+{"loss": 0.5757, "learning_rate": 8.170324161746685e-05, "epoch": 1.46, "step": 39310}
+{"loss": 0.5595, "learning_rate": 8.164754372284726e-05, "epoch": 1.46, "step": 39320}
+{"loss": 0.5609, "learning_rate": 8.159184582822769e-05, "epoch": 1.46, "step": 39330}
+{"loss": 0.6168, "learning_rate": 8.153614793360811e-05, "epoch": 1.46, "step": 39340}
+{"loss": 0.5133, "learning_rate": 8.148045003898852e-05, "epoch": 1.46, "step": 39350}
+{"loss": 0.6391, "learning_rate": 8.142475214436893e-05, "epoch": 1.46, "step": 39360}
+{"loss": 0.4574, "learning_rate": 8.136905424974936e-05, "epoch": 1.46, "step": 39370}
+{"loss": 0.7241, "learning_rate": 8.131335635512977e-05, "epoch": 1.46, "step": 39380}
+{"loss": 0.5844, "learning_rate": 8.125765846051018e-05, "epoch": 1.46, "step": 39390}
+{"loss": 0.6103, "learning_rate": 8.120196056589059e-05, "epoch": 1.46, "step": 39400}
+{"loss": 0.5897, "learning_rate": 8.114626267127102e-05, "epoch": 1.46, "step": 39410}
+{"loss": 0.5035, "learning_rate": 8.109056477665143e-05, "epoch": 1.46, "step": 39420}
+{"loss": 0.5493, "learning_rate": 8.103486688203184e-05, "epoch": 1.46, "step": 39430}
+{"loss": 0.478, "learning_rate": 8.097916898741228e-05, "epoch": 1.46, "step": 39440}
+{"loss": 0.5897, "learning_rate": 8.092347109279269e-05, "epoch": 1.46, "step": 39450}
+{"loss": 0.5199, "learning_rate": 8.08677731981731e-05, "epoch": 1.46, "step": 39460}
+{"loss": 0.6233, "learning_rate": 8.081207530355352e-05, "epoch": 1.46, "step": 39470}
+{"loss": 0.6201, "learning_rate": 8.075637740893394e-05, "epoch": 1.46, "step": 39480}
+{"loss": 0.7711, "learning_rate": 8.070067951431435e-05, "epoch": 1.46, "step": 39490}
+{"loss": 0.5393, "learning_rate": 8.064498161969477e-05, "epoch": 1.46, "step": 39500}
+{"loss": 0.6425, "learning_rate": 8.058928372507518e-05, "epoch": 1.46, "step": 39510}
+{"loss": 0.4941, "learning_rate": 8.05335858304556e-05, "epoch": 1.46, "step": 39520}
+{"loss": 0.6082, "learning_rate": 8.0477887935836e-05, "epoch": 1.47, "step": 39530}
+{"loss": 0.5083, "learning_rate": 8.042219004121644e-05, "epoch": 1.47, "step": 39540}
+{"loss": 0.5724, "learning_rate": 8.036649214659685e-05, "epoch": 1.47, "step": 39550}
+{"loss": 0.5116, "learning_rate": 8.031079425197727e-05, "epoch": 1.47, "step": 39560}
+{"loss": 0.6118, "learning_rate": 8.025509635735769e-05, "epoch": 1.47, "step": 39570}
+{"loss": 0.5039, "learning_rate": 8.01993984627381e-05, "epoch": 1.47, "step": 39580}
+{"loss": 0.5212, "learning_rate": 8.014370056811851e-05, "epoch": 1.47, "step": 39590}
+{"loss": 0.5142, "learning_rate": 8.008800267349894e-05, "epoch": 1.47, "step": 39600}
+{"loss": 0.575, "learning_rate": 8.003230477887935e-05, "epoch": 1.47, "step": 39610}
+{"loss": 0.5748, "learning_rate": 7.997660688425976e-05, "epoch": 1.47, "step": 39620}
+{"loss": 0.5913, "learning_rate": 7.99209089896402e-05, "epoch": 1.47, "step": 39630}
+{"loss": 0.5368, "learning_rate": 7.986521109502061e-05, "epoch": 1.47, "step": 39640}
+{"loss": 0.6079, "learning_rate": 7.980951320040102e-05, "epoch": 1.47, "step": 39650}
+{"loss": 0.6189, "learning_rate": 7.975381530578143e-05, "epoch": 1.47, "step": 39660}
+{"loss": 0.5868, "learning_rate": 7.969811741116186e-05, "epoch": 1.47, "step": 39670}
+{"loss": 0.551, "learning_rate": 7.964241951654227e-05, "epoch": 1.47, "step": 39680}
+{"loss": 0.5548, "learning_rate": 7.958672162192268e-05, "epoch": 1.47, "step": 39690}
+{"loss": 0.6888, "learning_rate": 7.95310237273031e-05, "epoch": 1.47, "step": 39700}
+{"loss": 0.6626, "learning_rate": 7.947532583268351e-05, "epoch": 1.47, "step": 39710}
+{"loss": 0.5973, "learning_rate": 7.941962793806393e-05, "epoch": 1.47, "step": 39720}
+{"loss": 0.5248, "learning_rate": 7.936393004344436e-05, "epoch": 1.47, "step": 39730}
+{"loss": 0.5731, "learning_rate": 7.930823214882477e-05, "epoch": 1.47, "step": 39740}
+{"loss": 0.7173, "learning_rate": 7.925253425420519e-05, "epoch": 1.47, "step": 39750}
+{"loss": 0.4956, "learning_rate": 7.919683635958561e-05, "epoch": 1.47, "step": 39760}
+{"loss": 0.6819, "learning_rate": 7.914113846496602e-05, "epoch": 1.47, "step": 39770}
+{"loss": 0.5895, "learning_rate": 7.908544057034643e-05, "epoch": 1.47, "step": 39780}
+{"loss": 0.6268, "learning_rate": 7.902974267572684e-05, "epoch": 1.47, "step": 39790}
+{"loss": 0.6219, "learning_rate": 7.897404478110727e-05, "epoch": 1.48, "step": 39800}
+{"loss": 0.6172, "learning_rate": 7.891834688648768e-05, "epoch": 1.48, "step": 39810}
+{"loss": 0.6433, "learning_rate": 7.886264899186809e-05, "epoch": 1.48, "step": 39820}
+{"loss": 0.597, "learning_rate": 7.880695109724852e-05, "epoch": 1.48, "step": 39830}
+{"loss": 0.5338, "learning_rate": 7.875125320262893e-05, "epoch": 1.48, "step": 39840}
+{"loss": 0.6401, "learning_rate": 7.869555530800935e-05, "epoch": 1.48, "step": 39850}
+{"loss": 0.5635, "learning_rate": 7.863985741338978e-05, "epoch": 1.48, "step": 39860}
+{"loss": 0.6666, "learning_rate": 7.858415951877019e-05, "epoch": 1.48, "step": 39870}
+{"loss": 0.5134, "learning_rate": 7.85284616241506e-05, "epoch": 1.48, "step": 39880}
+{"loss": 0.7161, "learning_rate": 7.847276372953102e-05, "epoch": 1.48, "step": 39890}
+{"loss": 0.4848, "learning_rate": 7.841706583491143e-05, "epoch": 1.48, "step": 39900}
+{"loss": 0.7374, "learning_rate": 7.836136794029185e-05, "epoch": 1.48, "step": 39910}
+{"loss": 0.5928, "learning_rate": 7.830567004567226e-05, "epoch": 1.48, "step": 39920}
+{"loss": 0.6739, "learning_rate": 7.824997215105268e-05, "epoch": 1.48, "step": 39930}
+{"loss": 0.5335, "learning_rate": 7.819427425643309e-05, "epoch": 1.48, "step": 39940}
+{"loss": 0.4856, "learning_rate": 7.81385763618135e-05, "epoch": 1.48, "step": 39950}
+{"loss": 0.6146, "learning_rate": 7.808287846719394e-05, "epoch": 1.48, "step": 39960}
+{"loss": 0.6421, "learning_rate": 7.802718057257435e-05, "epoch": 1.48, "step": 39970}
+{"loss": 0.503, "learning_rate": 7.797148267795476e-05, "epoch": 1.48, "step": 39980}
+{"loss": 0.5727, "learning_rate": 7.791578478333519e-05, "epoch": 1.48, "step": 39990}
+{"loss": 0.4706, "learning_rate": 7.78600868887156e-05, "epoch": 1.48, "step": 40000}
+{"loss": 0.4936, "learning_rate": 7.780438899409601e-05, "epoch": 1.48, "step": 40010}
+{"loss": 0.6032, "learning_rate": 7.774869109947644e-05, "epoch": 1.48, "step": 40020}
+{"loss": 0.5765, "learning_rate": 7.769299320485685e-05, "epoch": 1.48, "step": 40030}
+{"loss": 0.5574, "learning_rate": 7.763729531023726e-05, "epoch": 1.48, "step": 40040}
+{"loss": 0.6039, "learning_rate": 7.75815974156177e-05, "epoch": 1.48, "step": 40050}
+{"loss": 0.5606, "learning_rate": 7.752589952099811e-05, "epoch": 1.48, "step": 40060}
+{"loss": 0.5112, "learning_rate": 7.747020162637852e-05, "epoch": 1.49, "step": 40070}
+{"loss": 0.5822, "learning_rate": 7.741450373175893e-05, "epoch": 1.49, "step": 40080}
+{"loss": 0.7732, "learning_rate": 7.735880583713935e-05, "epoch": 1.49, "step": 40090}
+{"loss": 0.5575, "learning_rate": 7.730310794251977e-05, "epoch": 1.49, "step": 40100}
+{"loss": 0.6468, "learning_rate": 7.724741004790018e-05, "epoch": 1.49, "step": 40110}
+{"loss": 0.5499, "learning_rate": 7.71917121532806e-05, "epoch": 1.49, "step": 40120}
+{"loss": 0.6773, "learning_rate": 7.713601425866101e-05, "epoch": 1.49, "step": 40130}
+{"loss": 0.6287, "learning_rate": 7.708031636404142e-05, "epoch": 1.49, "step": 40140}
+{"loss": 0.4615, "learning_rate": 7.702461846942186e-05, "epoch": 1.49, "step": 40150}
+{"loss": 0.5807, "learning_rate": 7.696892057480227e-05, "epoch": 1.49, "step": 40160}
+{"loss": 0.5307, "learning_rate": 7.691322268018268e-05, "epoch": 1.49, "step": 40170}
+{"loss": 0.6332, "learning_rate": 7.685752478556311e-05, "epoch": 1.49, "step": 40180}
+{"loss": 0.4301, "learning_rate": 7.680182689094352e-05, "epoch": 1.49, "step": 40190}
+{"loss": 0.439, "learning_rate": 7.674612899632393e-05, "epoch": 1.49, "step": 40200}
+{"loss": 0.5922, "learning_rate": 7.669043110170434e-05, "epoch": 1.49, "step": 40210}
+{"loss": 0.5614, "learning_rate": 7.663473320708477e-05, "epoch": 1.49, "step": 40220}
+{"loss": 0.552, "learning_rate": 7.658460510192714e-05, "epoch": 1.49, "step": 40230}
+{"loss": 0.422, "learning_rate": 7.652890720730755e-05, "epoch": 1.49, "step": 40240}
+{"loss": 0.6389, "learning_rate": 7.647320931268796e-05, "epoch": 1.49, "step": 40250}
+{"loss": 0.5084, "learning_rate": 7.64175114180684e-05, "epoch": 1.49, "step": 40260}
+{"loss": 0.5967, "learning_rate": 7.636181352344881e-05, "epoch": 1.49, "step": 40270}
+{"loss": 0.6254, "learning_rate": 7.630611562882922e-05, "epoch": 1.49, "step": 40280}
+{"loss": 0.6903, "learning_rate": 7.625041773420965e-05, "epoch": 1.49, "step": 40290}
+{"loss": 0.5762, "learning_rate": 7.619471983959006e-05, "epoch": 1.49, "step": 40300}
+{"loss": 0.4594, "learning_rate": 7.613902194497047e-05, "epoch": 1.49, "step": 40310}
+{"loss": 0.5053, "learning_rate": 7.60833240503509e-05, "epoch": 1.49, "step": 40320}
+{"loss": 0.5539, "learning_rate": 7.60276261557313e-05, "epoch": 1.49, "step": 40330}
+{"loss": 0.6412, "learning_rate": 7.597192826111172e-05, "epoch": 1.5, "step": 40340}
+{"loss": 0.7587, "learning_rate": 7.591623036649215e-05, "epoch": 1.5, "step": 40350}
+{"loss": 0.6224, "learning_rate": 7.586053247187257e-05, "epoch": 1.5, "step": 40360}
+{"loss": 0.5366, "learning_rate": 7.580483457725298e-05, "epoch": 1.5, "step": 40370}
+{"loss": 0.4758, "learning_rate": 7.574913668263339e-05, "epoch": 1.5, "step": 40380}
+{"loss": 0.5738, "learning_rate": 7.569343878801381e-05, "epoch": 1.5, "step": 40390}
+{"loss": 0.6283, "learning_rate": 7.563774089339422e-05, "epoch": 1.5, "step": 40400}
+{"loss": 0.7389, "learning_rate": 7.558204299877463e-05, "epoch": 1.5, "step": 40410}
+{"loss": 0.5932, "learning_rate": 7.552634510415506e-05, "epoch": 1.5, "step": 40420}
+{"loss": 0.5267, "learning_rate": 7.547064720953547e-05, "epoch": 1.5, "step": 40430}
+{"loss": 0.6217, "learning_rate": 7.541494931491588e-05, "epoch": 1.5, "step": 40440}
+{"loss": 0.8216, "learning_rate": 7.53592514202963e-05, "epoch": 1.5, "step": 40450}
+{"loss": 0.5606, "learning_rate": 7.530355352567672e-05, "epoch": 1.5, "step": 40460}
+{"loss": 0.4766, "learning_rate": 7.524785563105714e-05, "epoch": 1.5, "step": 40470}
+{"loss": 0.5511, "learning_rate": 7.519215773643757e-05, "epoch": 1.5, "step": 40480}
+{"loss": 0.6062, "learning_rate": 7.513645984181798e-05, "epoch": 1.5, "step": 40490}
+{"loss": 0.621, "learning_rate": 7.508076194719839e-05, "epoch": 1.5, "step": 40500}
+{"loss": 0.5645, "learning_rate": 7.50250640525788e-05, "epoch": 1.5, "step": 40510}
+{"loss": 0.6488, "learning_rate": 7.496936615795922e-05, "epoch": 1.5, "step": 40520}
+{"loss": 0.5186, "learning_rate": 7.491366826333964e-05, "epoch": 1.5, "step": 40530}
+{"loss": 0.5323, "learning_rate": 7.485797036872006e-05, "epoch": 1.5, "step": 40540}
+{"loss": 0.6308, "learning_rate": 7.480227247410047e-05, "epoch": 1.5, "step": 40550}
+{"loss": 0.5674, "learning_rate": 7.474657457948088e-05, "epoch": 1.5, "step": 40560}
+{"loss": 0.6205, "learning_rate": 7.469087668486131e-05, "epoch": 1.5, "step": 40570}
+{"loss": 0.5036, "learning_rate": 7.463517879024172e-05, "epoch": 1.5, "step": 40580}
+{"loss": 0.5233, "learning_rate": 7.457948089562214e-05, "epoch": 1.5, "step": 40590}
+{"loss": 0.6184, "learning_rate": 7.452378300100255e-05, "epoch": 1.5, "step": 40600}
+{"loss": 0.5887, "learning_rate": 7.446808510638297e-05, "epoch": 1.51, "step": 40610}
+{"loss": 0.4676, "learning_rate": 7.441238721176339e-05, "epoch": 1.51, "step": 40620}
+{"loss": 0.4686, "learning_rate": 7.43566893171438e-05, "epoch": 1.51, "step": 40630}
+{"loss": 0.5638, "learning_rate": 7.430099142252423e-05, "epoch": 1.51, "step": 40640}
+{"loss": 0.5497, "learning_rate": 7.424529352790464e-05, "epoch": 1.51, "step": 40650}
+{"loss": 0.6577, "learning_rate": 7.418959563328505e-05, "epoch": 1.51, "step": 40660}
+{"loss": 0.5928, "learning_rate": 7.413389773866547e-05, "epoch": 1.51, "step": 40670}
+{"loss": 0.6367, "learning_rate": 7.407819984404588e-05, "epoch": 1.51, "step": 40680}
+{"loss": 0.5934, "learning_rate": 7.402250194942631e-05, "epoch": 1.51, "step": 40690}
+{"loss": 0.5527, "learning_rate": 7.396680405480672e-05, "epoch": 1.51, "step": 40700}
+{"loss": 0.6522, "learning_rate": 7.391110616018713e-05, "epoch": 1.51, "step": 40710}
+{"loss": 0.5563, "learning_rate": 7.385540826556756e-05, "epoch": 1.51, "step": 40720}
+{"loss": 0.6284, "learning_rate": 7.379971037094797e-05, "epoch": 1.51, "step": 40730}
+{"loss": 0.4895, "learning_rate": 7.374401247632839e-05, "epoch": 1.51, "step": 40740}
+{"loss": 0.5064, "learning_rate": 7.36883145817088e-05, "epoch": 1.51, "step": 40750}
+{"loss": 0.5851, "learning_rate": 7.363261668708921e-05, "epoch": 1.51, "step": 40760}
+{"loss": 0.6601, "learning_rate": 7.357691879246964e-05, "epoch": 1.51, "step": 40770}
+{"loss": 0.6701, "learning_rate": 7.352122089785006e-05, "epoch": 1.51, "step": 40780}
+{"loss": 0.464, "learning_rate": 7.346552300323047e-05, "epoch": 1.51, "step": 40790}
+{"loss": 0.6026, "learning_rate": 7.340982510861089e-05, "epoch": 1.51, "step": 40800}
+{"loss": 0.6416, "learning_rate": 7.33541272139913e-05, "epoch": 1.51, "step": 40810}
+{"loss": 0.5212, "learning_rate": 7.329842931937172e-05, "epoch": 1.51, "step": 40820}
+{"loss": 0.6097, "learning_rate": 7.324273142475215e-05, "epoch": 1.51, "step": 40830}
+{"loss": 0.5378, "learning_rate": 7.318703353013256e-05, "epoch": 1.51, "step": 40840}
+{"loss": 0.5055, "learning_rate": 7.313133563551297e-05, "epoch": 1.51, "step": 40850}
+{"loss": 0.5933, "learning_rate": 7.30756377408934e-05, "epoch": 1.51, "step": 40860}
+{"loss": 0.6016, "learning_rate": 7.30199398462738e-05, "epoch": 1.51, "step": 40870}
+{"loss": 0.6743, "learning_rate": 7.296424195165423e-05, "epoch": 1.52, "step": 40880}
+{"loss": 0.6074, "learning_rate": 7.290854405703464e-05, "epoch": 1.52, "step": 40890}
+{"loss": 0.6888, "learning_rate": 7.285284616241505e-05, "epoch": 1.52, "step": 40900}
+{"loss": 0.6754, "learning_rate": 7.279714826779548e-05, "epoch": 1.52, "step": 40910}
+{"loss": 0.5157, "learning_rate": 7.274145037317589e-05, "epoch": 1.52, "step": 40920}
+{"loss": 0.5088, "learning_rate": 7.268575247855631e-05, "epoch": 1.52, "step": 40930}
+{"loss": 0.5971, "learning_rate": 7.263005458393672e-05, "epoch": 1.52, "step": 40940}
+{"loss": 0.573, "learning_rate": 7.257435668931713e-05, "epoch": 1.52, "step": 40950}
+{"loss": 0.6144, "learning_rate": 7.251865879469756e-05, "epoch": 1.52, "step": 40960}
+{"loss": 0.6003, "learning_rate": 7.246296090007797e-05, "epoch": 1.52, "step": 40970}
+{"loss": 0.5963, "learning_rate": 7.24072630054584e-05, "epoch": 1.52, "step": 40980}
+{"loss": 0.6157, "learning_rate": 7.23515651108388e-05, "epoch": 1.52, "step": 40990}
+{"loss": 0.6737, "learning_rate": 7.229586721621922e-05, "epoch": 1.52, "step": 41000}
+{"loss": 0.5582, "learning_rate": 7.224016932159964e-05, "epoch": 1.52, "step": 41010}
+{"loss": 0.5435, "learning_rate": 7.218447142698005e-05, "epoch": 1.52, "step": 41020}
+{"loss": 0.5482, "learning_rate": 7.212877353236046e-05, "epoch": 1.52, "step": 41030}
+{"loss": 0.5183, "learning_rate": 7.207307563774089e-05, "epoch": 1.52, "step": 41040}
+{"loss": 0.6322, "learning_rate": 7.20173777431213e-05, "epoch": 1.52, "step": 41050}
+{"loss": 0.5983, "learning_rate": 7.196167984850172e-05, "epoch": 1.52, "step": 41060}
+{"loss": 0.4881, "learning_rate": 7.190598195388214e-05, "epoch": 1.52, "step": 41070}
+{"loss": 0.6044, "learning_rate": 7.185028405926255e-05, "epoch": 1.52, "step": 41080}
+{"loss": 0.4251, "learning_rate": 7.179458616464297e-05, "epoch": 1.52, "step": 41090}
+{"loss": 0.6917, "learning_rate": 7.173888827002338e-05, "epoch": 1.52, "step": 41100}
+{"loss": 0.6223, "learning_rate": 7.168319037540381e-05, "epoch": 1.52, "step": 41110}
+{"loss": 0.536, "learning_rate": 7.162749248078422e-05, "epoch": 1.52, "step": 41120}
+{"loss": 0.6086, "learning_rate": 7.157179458616463e-05, "epoch": 1.52, "step": 41130}
+{"loss": 0.5963, "learning_rate": 7.151609669154505e-05, "epoch": 1.52, "step": 41140}
+{"loss": 0.5298, "learning_rate": 7.146039879692547e-05, "epoch": 1.53, "step": 41150}
+{"loss": 0.6451, "learning_rate": 7.140470090230589e-05, "epoch": 1.53, "step": 41160}
+{"loss": 0.7617, "learning_rate": 7.13490030076863e-05, "epoch": 1.53, "step": 41170}
+{"loss": 0.6068, "learning_rate": 7.129330511306671e-05, "epoch": 1.53, "step": 41180}
+{"loss": 0.5597, "learning_rate": 7.123760721844714e-05, "epoch": 1.53, "step": 41190}
+{"loss": 0.6196, "learning_rate": 7.118190932382755e-05, "epoch": 1.53, "step": 41200}
+{"loss": 0.5273, "learning_rate": 7.112621142920797e-05, "epoch": 1.53, "step": 41210}
+{"loss": 0.431, "learning_rate": 7.107051353458838e-05, "epoch": 1.53, "step": 41220}
+{"loss": 0.587, "learning_rate": 7.10148156399688e-05, "epoch": 1.53, "step": 41230}
+{"loss": 0.5756, "learning_rate": 7.095911774534922e-05, "epoch": 1.53, "step": 41240}
+{"loss": 0.4591, "learning_rate": 7.090341985072964e-05, "epoch": 1.53, "step": 41250}
+{"loss": 0.447, "learning_rate": 7.084772195611006e-05, "epoch": 1.53, "step": 41260}
+{"loss": 0.5481, "learning_rate": 7.079202406149047e-05, "epoch": 1.53, "step": 41270}
+{"loss": 0.5159, "learning_rate": 7.073632616687088e-05, "epoch": 1.53, "step": 41280}
+{"loss": 0.5976, "learning_rate": 7.06806282722513e-05, "epoch": 1.53, "step": 41290}
+{"loss": 0.4891, "learning_rate": 7.062493037763173e-05, "epoch": 1.53, "step": 41300}
+{"loss": 0.694, "learning_rate": 7.056923248301214e-05, "epoch": 1.53, "step": 41310}
+{"loss": 0.5772, "learning_rate": 7.051353458839255e-05, "epoch": 1.53, "step": 41320}
+{"loss": 0.6043, "learning_rate": 7.045783669377297e-05, "epoch": 1.53, "step": 41330}
+{"loss": 0.6163, "learning_rate": 7.040213879915339e-05, "epoch": 1.53, "step": 41340}
+{"loss": 0.567, "learning_rate": 7.034644090453381e-05, "epoch": 1.53, "step": 41350}
+{"loss": 0.5425, "learning_rate": 7.029074300991422e-05, "epoch": 1.53, "step": 41360}
+{"loss": 0.6521, "learning_rate": 7.023504511529463e-05, "epoch": 1.53, "step": 41370}
+{"loss": 0.7866, "learning_rate": 7.017934722067506e-05, "epoch": 1.53, "step": 41380}
+{"loss": 0.5658, "learning_rate": 7.012364932605547e-05, "epoch": 1.53, "step": 41390}
+{"loss": 0.5146, "learning_rate": 7.006795143143589e-05, "epoch": 1.53, "step": 41400}
+{"loss": 0.5498, "learning_rate": 7.00122535368163e-05, "epoch": 1.53, "step": 41410}
+{"loss": 0.6435, "learning_rate": 6.995655564219672e-05, "epoch": 1.54, "step": 41420}
+{"loss": 0.4407, "learning_rate": 6.990085774757714e-05, "epoch": 1.54, "step": 41430}
+{"loss": 0.5896, "learning_rate": 6.984515985295755e-05, "epoch": 1.54, "step": 41440}
+{"loss": 0.6194, "learning_rate": 6.978946195833798e-05, "epoch": 1.54, "step": 41450}
+{"loss": 0.5176, "learning_rate": 6.973376406371839e-05, "epoch": 1.54, "step": 41460}
+{"loss": 0.6486, "learning_rate": 6.96780661690988e-05, "epoch": 1.54, "step": 41470}
+{"loss": 0.7138, "learning_rate": 6.962236827447922e-05, "epoch": 1.54, "step": 41480}
+{"loss": 0.6464, "learning_rate": 6.956667037985963e-05, "epoch": 1.54, "step": 41490}
+{"loss": 0.5833, "learning_rate": 6.951097248524006e-05, "epoch": 1.54, "step": 41500}
+{"loss": 0.56, "learning_rate": 6.945527459062047e-05, "epoch": 1.54, "step": 41510}
+{"loss": 0.4471, "learning_rate": 6.939957669600088e-05, "epoch": 1.54, "step": 41520}
+{"loss": 0.5446, "learning_rate": 6.93438788013813e-05, "epoch": 1.54, "step": 41530}
+{"loss": 0.4813, "learning_rate": 6.928818090676172e-05, "epoch": 1.54, "step": 41540}
+{"loss": 0.6535, "learning_rate": 6.923248301214213e-05, "epoch": 1.54, "step": 41550}
+{"loss": 0.525, "learning_rate": 6.917678511752255e-05, "epoch": 1.54, "step": 41560}
+{"loss": 0.6113, "learning_rate": 6.912108722290296e-05, "epoch": 1.54, "step": 41570}
+{"loss": 0.5219, "learning_rate": 6.906538932828339e-05, "epoch": 1.54, "step": 41580}
+{"loss": 0.6028, "learning_rate": 6.90096914336638e-05, "epoch": 1.54, "step": 41590}
+{"loss": 0.4393, "learning_rate": 6.895399353904421e-05, "epoch": 1.54, "step": 41600}
+{"loss": 0.575, "learning_rate": 6.889829564442464e-05, "epoch": 1.54, "step": 41610}
+{"loss": 0.6329, "learning_rate": 6.884259774980505e-05, "epoch": 1.54, "step": 41620}
+{"loss": 0.6352, "learning_rate": 6.878689985518547e-05, "epoch": 1.54, "step": 41630}
+{"loss": 0.5936, "learning_rate": 6.873120196056588e-05, "epoch": 1.54, "step": 41640}
+{"loss": 0.471, "learning_rate": 6.867550406594629e-05, "epoch": 1.54, "step": 41650}
+{"loss": 0.6303, "learning_rate": 6.861980617132672e-05, "epoch": 1.54, "step": 41660}
+{"loss": 0.5688, "learning_rate": 6.856410827670713e-05, "epoch": 1.54, "step": 41670}
+{"loss": 0.4856, "learning_rate": 6.850841038208755e-05, "epoch": 1.54, "step": 41680}
+{"loss": 0.5973, "learning_rate": 6.845271248746796e-05, "epoch": 1.55, "step": 41690}
+{"loss": 0.5696, "learning_rate": 6.839701459284838e-05, "epoch": 1.55, "step": 41700}
+{"loss": 0.5856, "learning_rate": 6.83413166982288e-05, "epoch": 1.55, "step": 41710}
+{"loss": 0.7105, "learning_rate": 6.828561880360923e-05, "epoch": 1.55, "step": 41720}
+{"loss": 0.6047, "learning_rate": 6.822992090898964e-05, "epoch": 1.55, "step": 41730}
+{"loss": 0.5922, "learning_rate": 6.817422301437005e-05, "epoch": 1.55, "step": 41740}
+{"loss": 0.5163, "learning_rate": 6.811852511975046e-05, "epoch": 1.55, "step": 41750}
+{"loss": 0.5799, "learning_rate": 6.806282722513088e-05, "epoch": 1.55, "step": 41760}
+{"loss": 0.6015, "learning_rate": 6.800712933051131e-05, "epoch": 1.55, "step": 41770}
+{"loss": 0.6168, "learning_rate": 6.795143143589172e-05, "epoch": 1.55, "step": 41780}
+{"loss": 0.546, "learning_rate": 6.789573354127213e-05, "epoch": 1.55, "step": 41790}
+{"loss": 0.571, "learning_rate": 6.784003564665256e-05, "epoch": 1.55, "step": 41800}
+{"loss": 0.4898, "learning_rate": 6.778433775203297e-05, "epoch": 1.55, "step": 41810}
+{"loss": 0.5061, "learning_rate": 6.772863985741339e-05, "epoch": 1.55, "step": 41820}
+{"loss": 0.5205, "learning_rate": 6.76729419627938e-05, "epoch": 1.55, "step": 41830}
+{"loss": 0.5441, "learning_rate": 6.761724406817421e-05, "epoch": 1.55, "step": 41840}
+{"loss": 0.5502, "learning_rate": 6.756154617355464e-05, "epoch": 1.55, "step": 41850}
+{"loss": 0.5188, "learning_rate": 6.750584827893505e-05, "epoch": 1.55, "step": 41860}
+{"loss": 0.5865, "learning_rate": 6.745015038431547e-05, "epoch": 1.55, "step": 41870}
+{"loss": 0.48, "learning_rate": 6.739445248969588e-05, "epoch": 1.55, "step": 41880}
+{"loss": 0.5491, "learning_rate": 6.73387545950763e-05, "epoch": 1.55, "step": 41890}
+{"loss": 0.5788, "learning_rate": 6.728305670045672e-05, "epoch": 1.55, "step": 41900}
+{"loss": 0.5824, "learning_rate": 6.722735880583713e-05, "epoch": 1.55, "step": 41910}
+{"loss": 0.5147, "learning_rate": 6.717166091121756e-05, "epoch": 1.55, "step": 41920}
+{"loss": 0.6484, "learning_rate": 6.711596301659797e-05, "epoch": 1.55, "step": 41930}
+{"loss": 0.429, "learning_rate": 6.706026512197838e-05, "epoch": 1.55, "step": 41940}
+{"loss": 0.6061, "learning_rate": 6.70045672273588e-05, "epoch": 1.55, "step": 41950}
+{"loss": 0.6317, "learning_rate": 6.694886933273921e-05, "epoch": 1.56, "step": 41960}
+{"loss": 0.5021, "learning_rate": 6.689317143811964e-05, "epoch": 1.56, "step": 41970}
+{"loss": 0.6281, "learning_rate": 6.683747354350005e-05, "epoch": 1.56, "step": 41980}
+{"loss": 0.5148, "learning_rate": 6.678177564888046e-05, "epoch": 1.56, "step": 41990}
+{"loss": 0.584, "learning_rate": 6.672607775426089e-05, "epoch": 1.56, "step": 42000}
+{"loss": 0.6127, "learning_rate": 6.66703798596413e-05, "epoch": 1.56, "step": 42010}
+{"loss": 0.6296, "learning_rate": 6.661468196502172e-05, "epoch": 1.56, "step": 42020}
+{"loss": 0.5686, "learning_rate": 6.655898407040213e-05, "epoch": 1.56, "step": 42030}
+{"loss": 0.5753, "learning_rate": 6.650328617578254e-05, "epoch": 1.56, "step": 42040}
+{"loss": 0.4686, "learning_rate": 6.644758828116297e-05, "epoch": 1.56, "step": 42050}
+{"loss": 0.7022, "learning_rate": 6.639189038654338e-05, "epoch": 1.56, "step": 42060}
+{"loss": 0.6933, "learning_rate": 6.633619249192379e-05, "epoch": 1.56, "step": 42070}
+{"loss": 0.5129, "learning_rate": 6.628049459730422e-05, "epoch": 1.56, "step": 42080}
+{"loss": 0.6067, "learning_rate": 6.622479670268463e-05, "epoch": 1.56, "step": 42090}
+{"loss": 0.6279, "learning_rate": 6.616909880806505e-05, "epoch": 1.56, "step": 42100}
+{"loss": 0.4647, "learning_rate": 6.611340091344546e-05, "epoch": 1.56, "step": 42110}
+{"loss": 0.4879, "learning_rate": 6.605770301882587e-05, "epoch": 1.56, "step": 42120}
+{"loss": 0.4727, "learning_rate": 6.60020051242063e-05, "epoch": 1.56, "step": 42130}
+{"loss": 0.5251, "learning_rate": 6.594630722958671e-05, "epoch": 1.56, "step": 42140}
+{"loss": 0.5472, "learning_rate": 6.589060933496713e-05, "epoch": 1.56, "step": 42150}
+{"loss": 0.6579, "learning_rate": 6.583491144034755e-05, "epoch": 1.56, "step": 42160}
+{"loss": 0.6365, "learning_rate": 6.577921354572796e-05, "epoch": 1.56, "step": 42170}
+{"loss": 0.5025, "learning_rate": 6.572351565110838e-05, "epoch": 1.56, "step": 42180}
+{"loss": 0.6741, "learning_rate": 6.56678177564888e-05, "epoch": 1.56, "step": 42190}
+{"loss": 0.5789, "learning_rate": 6.561211986186922e-05, "epoch": 1.56, "step": 42200}
+{"loss": 0.563, "learning_rate": 6.555642196724963e-05, "epoch": 1.56, "step": 42210}
+{"loss": 0.4188, "learning_rate": 6.550072407263004e-05, "epoch": 1.56, "step": 42220}
+{"loss": 0.4501, "learning_rate": 6.544502617801046e-05, "epoch": 1.57, "step": 42230}
+{"loss": 0.6077, "learning_rate": 6.538932828339089e-05, "epoch": 1.57, "step": 42240}
+{"loss": 0.448, "learning_rate": 6.53336303887713e-05, "epoch": 1.57, "step": 42250}
+{"loss": 0.5281, "learning_rate": 6.527793249415171e-05, "epoch": 1.57, "step": 42260}
+{"loss": 0.5279, "learning_rate": 6.522223459953214e-05, "epoch": 1.57, "step": 42270}
+{"loss": 0.587, "learning_rate": 6.516653670491255e-05, "epoch": 1.57, "step": 42280}
+{"loss": 0.6235, "learning_rate": 6.511083881029297e-05, "epoch": 1.57, "step": 42290}
+{"loss": 0.5993, "learning_rate": 6.505514091567338e-05, "epoch": 1.57, "step": 42300}
+{"loss": 0.491, "learning_rate": 6.49994430210538e-05, "epoch": 1.57, "step": 42310}
+{"loss": 0.627, "learning_rate": 6.494374512643422e-05, "epoch": 1.57, "step": 42320}
+{"loss": 0.545, "learning_rate": 6.488804723181463e-05, "epoch": 1.57, "step": 42330}
+{"loss": 0.71, "learning_rate": 6.483234933719505e-05, "epoch": 1.57, "step": 42340}
+{"loss": 0.5984, "learning_rate": 6.477665144257547e-05, "epoch": 1.57, "step": 42350}
+{"loss": 0.7135, "learning_rate": 6.472095354795588e-05, "epoch": 1.57, "step": 42360}
+{"loss": 0.4812, "learning_rate": 6.46652556533363e-05, "epoch": 1.57, "step": 42370}
+{"loss": 0.6753, "learning_rate": 6.460955775871671e-05, "epoch": 1.57, "step": 42380}
+{"loss": 0.5908, "learning_rate": 6.455385986409714e-05, "epoch": 1.57, "step": 42390}
+{"loss": 0.5926, "learning_rate": 6.449816196947755e-05, "epoch": 1.57, "step": 42400}
+{"loss": 0.4748, "learning_rate": 6.444246407485796e-05, "epoch": 1.57, "step": 42410}
+{"loss": 0.6817, "learning_rate": 6.438676618023838e-05, "epoch": 1.57, "step": 42420}
+{"loss": 0.4508, "learning_rate": 6.43310682856188e-05, "epoch": 1.57, "step": 42430}
+{"loss": 0.5497, "learning_rate": 6.427537039099922e-05, "epoch": 1.57, "step": 42440}
+{"loss": 0.5681, "learning_rate": 6.421967249637963e-05, "epoch": 1.57, "step": 42450}
+{"loss": 0.6295, "learning_rate": 6.416397460176004e-05, "epoch": 1.57, "step": 42460}
+{"loss": 0.4416, "learning_rate": 6.410827670714047e-05, "epoch": 1.57, "step": 42470}
+{"loss": 0.5661, "learning_rate": 6.405257881252088e-05, "epoch": 1.57, "step": 42480}
+{"loss": 0.5799, "learning_rate": 6.39968809179013e-05, "epoch": 1.57, "step": 42490}
+{"loss": 0.4385, "learning_rate": 6.394118302328171e-05, "epoch": 1.58, "step": 42500}
+{"loss": 0.6964, "learning_rate": 6.388548512866213e-05, "epoch": 1.58, "step": 42510}
+{"loss": 0.5238, "learning_rate": 6.382978723404255e-05, "epoch": 1.58, "step": 42520}
+{"loss": 0.5934, "learning_rate": 6.377408933942297e-05, "epoch": 1.58, "step": 42530}
+{"loss": 0.5819, "learning_rate": 6.371839144480339e-05, "epoch": 1.58, "step": 42540}
+{"loss": 0.5154, "learning_rate": 6.36626935501838e-05, "epoch": 1.58, "step": 42550}
+{"loss": 0.513, "learning_rate": 6.360699565556421e-05, "epoch": 1.58, "step": 42560}
+{"loss": 0.5352, "learning_rate": 6.355129776094463e-05, "epoch": 1.58, "step": 42570}
+{"loss": 0.6589, "learning_rate": 6.349559986632504e-05, "epoch": 1.58, "step": 42580}
+{"loss": 0.5297, "learning_rate": 6.343990197170547e-05, "epoch": 1.58, "step": 42590}
+{"loss": 0.6475, "learning_rate": 6.338420407708588e-05, "epoch": 1.58, "step": 42600}
+{"loss": 0.624, "learning_rate": 6.332850618246629e-05, "epoch": 1.58, "step": 42610}
+{"loss": 0.55, "learning_rate": 6.327280828784672e-05, "epoch": 1.58, "step": 42620}
+{"loss": 0.5857, "learning_rate": 6.321711039322713e-05, "epoch": 1.58, "step": 42630}
+{"loss": 0.6391, "learning_rate": 6.316141249860754e-05, "epoch": 1.58, "step": 42640}
+{"loss": 0.7421, "learning_rate": 6.310571460398796e-05, "epoch": 1.58, "step": 42650}
+{"loss": 0.7518, "learning_rate": 6.305001670936839e-05, "epoch": 1.58, "step": 42660}
+{"loss": 0.5515, "learning_rate": 6.29943188147488e-05, "epoch": 1.58, "step": 42670}
+{"loss": 0.6017, "learning_rate": 6.293862092012921e-05, "epoch": 1.58, "step": 42680}
+{"loss": 0.4212, "learning_rate": 6.288292302550962e-05, "epoch": 1.58, "step": 42690}
+{"loss": 0.449, "learning_rate": 6.282722513089005e-05, "epoch": 1.58, "step": 42700}
+{"loss": 0.5952, "learning_rate": 6.277152723627047e-05, "epoch": 1.58, "step": 42710}
+{"loss": 0.6312, "learning_rate": 6.271582934165088e-05, "epoch": 1.58, "step": 42720}
+{"loss": 0.5841, "learning_rate": 6.266013144703129e-05, "epoch": 1.58, "step": 42730}
+{"loss": 0.6086, "learning_rate": 6.260443355241172e-05, "epoch": 1.58, "step": 42740}
+{"loss": 0.6145, "learning_rate": 6.254873565779213e-05, "epoch": 1.58, "step": 42750}
+{"loss": 0.5464, "learning_rate": 6.249303776317255e-05, "epoch": 1.58, "step": 42760}
+{"loss": 0.586, "learning_rate": 6.243733986855296e-05, "epoch": 1.59, "step": 42770}
+{"loss": 0.4917, "learning_rate": 6.238164197393337e-05, "epoch": 1.59, "step": 42780}
+{"loss": 0.614, "learning_rate": 6.23259440793138e-05, "epoch": 1.59, "step": 42790}
+{"loss": 0.702, "learning_rate": 6.227024618469421e-05, "epoch": 1.59, "step": 42800}
+{"loss": 0.6493, "learning_rate": 6.221454829007464e-05, "epoch": 1.59, "step": 42810}
+{"loss": 0.6244, "learning_rate": 6.215885039545505e-05, "epoch": 1.59, "step": 42820}
+{"loss": 0.5299, "learning_rate": 6.210315250083546e-05, "epoch": 1.59, "step": 42830}
+{"loss": 0.5824, "learning_rate": 6.204745460621588e-05, "epoch": 1.59, "step": 42840}
+{"loss": 0.5691, "learning_rate": 6.19917567115963e-05, "epoch": 1.59, "step": 42850}
+{"loss": 0.5034, "learning_rate": 6.193605881697672e-05, "epoch": 1.59, "step": 42860}
+{"loss": 0.6444, "learning_rate": 6.188036092235713e-05, "epoch": 1.59, "step": 42870}
+{"loss": 0.6555, "learning_rate": 6.182466302773754e-05, "epoch": 1.59, "step": 42880}
+{"loss": 0.5187, "learning_rate": 6.176896513311797e-05, "epoch": 1.59, "step": 42890}
+{"loss": 0.5819, "learning_rate": 6.171326723849838e-05, "epoch": 1.59, "step": 42900}
+{"loss": 0.5772, "learning_rate": 6.16575693438788e-05, "epoch": 1.59, "step": 42910}
+{"loss": 0.5957, "learning_rate": 6.160187144925921e-05, "epoch": 1.59, "step": 42920}
+{"loss": 0.5934, "learning_rate": 6.154617355463962e-05, "epoch": 1.59, "step": 42930}
+{"loss": 0.5445, "learning_rate": 6.149047566002005e-05, "epoch": 1.59, "step": 42940}
+{"loss": 0.6011, "learning_rate": 6.143477776540046e-05, "epoch": 1.59, "step": 42950}
+{"loss": 0.5211, "learning_rate": 6.137907987078088e-05, "epoch": 1.59, "step": 42960}
+{"loss": 0.5328, "learning_rate": 6.13233819761613e-05, "epoch": 1.59, "step": 42970}
+{"loss": 0.5839, "learning_rate": 6.12676840815417e-05, "epoch": 1.59, "step": 42980}
+{"loss": 0.5628, "learning_rate": 6.121198618692213e-05, "epoch": 1.59, "step": 42990}
+{"loss": 0.5969, "learning_rate": 6.115628829230256e-05, "epoch": 1.59, "step": 43000}
+{"loss": 0.6406, "learning_rate": 6.110059039768297e-05, "epoch": 1.59, "step": 43010}
+{"loss": 0.4462, "learning_rate": 6.104489250306338e-05, "epoch": 1.59, "step": 43020}
+{"loss": 0.5414, "learning_rate": 6.098919460844379e-05, "epoch": 1.59, "step": 43030}
+{"loss": 0.742, "learning_rate": 6.0933496713824214e-05, "epoch": 1.6, "step": 43040}
+{"loss": 0.5354, "learning_rate": 6.087779881920463e-05, "epoch": 1.6, "step": 43050}
+{"loss": 0.5085, "learning_rate": 6.082210092458504e-05, "epoch": 1.6, "step": 43060}
+{"loss": 0.4648, "learning_rate": 6.076640302996546e-05, "epoch": 1.6, "step": 43070}
+{"loss": 0.5113, "learning_rate": 6.071070513534587e-05, "epoch": 1.6, "step": 43080}
+{"loss": 0.723, "learning_rate": 6.0655007240726296e-05, "epoch": 1.6, "step": 43090}
+{"loss": 0.6523, "learning_rate": 6.0599309346106714e-05, "epoch": 1.6, "step": 43100}
+{"loss": 0.565, "learning_rate": 6.0543611451487125e-05, "epoch": 1.6, "step": 43110}
+{"loss": 0.5951, "learning_rate": 6.048791355686754e-05, "epoch": 1.6, "step": 43120}
+{"loss": 0.5108, "learning_rate": 6.043221566224797e-05, "epoch": 1.6, "step": 43130}
+{"loss": 0.5616, "learning_rate": 6.037651776762838e-05, "epoch": 1.6, "step": 43140}
+{"loss": 0.4755, "learning_rate": 6.03208198730088e-05, "epoch": 1.6, "step": 43150}
+{"loss": 0.5337, "learning_rate": 6.026512197838921e-05, "epoch": 1.6, "step": 43160}
+{"loss": 0.5642, "learning_rate": 6.0209424083769626e-05, "epoch": 1.6, "step": 43170}
+{"loss": 0.5967, "learning_rate": 6.015372618915005e-05, "epoch": 1.6, "step": 43180}
+{"loss": 0.5457, "learning_rate": 6.009802829453046e-05, "epoch": 1.6, "step": 43190}
+{"loss": 0.614, "learning_rate": 6.004233039991088e-05, "epoch": 1.6, "step": 43200}
+{"loss": 0.6379, "learning_rate": 5.99866325052913e-05, "epoch": 1.6, "step": 43210}
+{"loss": 0.4907, "learning_rate": 5.993093461067171e-05, "epoch": 1.6, "step": 43220}
+{"loss": 0.586, "learning_rate": 5.987523671605213e-05, "epoch": 1.6, "step": 43230}
+{"loss": 0.7276, "learning_rate": 5.9819538821432545e-05, "epoch": 1.6, "step": 43240}
+{"loss": 0.6253, "learning_rate": 5.976384092681296e-05, "epoch": 1.6, "step": 43250}
+{"loss": 0.5205, "learning_rate": 5.970814303219338e-05, "epoch": 1.6, "step": 43260}
+{"loss": 0.6194, "learning_rate": 5.965244513757379e-05, "epoch": 1.6, "step": 43270}
+{"loss": 0.537, "learning_rate": 5.959674724295421e-05, "epoch": 1.6, "step": 43280}
+{"loss": 0.4461, "learning_rate": 5.954104934833462e-05, "epoch": 1.6, "step": 43290}
+{"loss": 0.6683, "learning_rate": 5.9485351453715045e-05, "epoch": 1.6, "step": 43300}
+{"loss": 0.5558, "learning_rate": 5.942965355909546e-05, "epoch": 1.61, "step": 43310}
+{"loss": 0.7098, "learning_rate": 5.9373955664475874e-05, "epoch": 1.61, "step": 43320}
+{"loss": 0.5677, "learning_rate": 5.931825776985629e-05, "epoch": 1.61, "step": 43330}
+{"loss": 0.5365, "learning_rate": 5.926255987523672e-05, "epoch": 1.61, "step": 43340}
+{"loss": 0.4692, "learning_rate": 5.920686198061713e-05, "epoch": 1.61, "step": 43350}
+{"loss": 0.6152, "learning_rate": 5.9151164085997546e-05, "epoch": 1.61, "step": 43360}
+{"loss": 0.5987, "learning_rate": 5.909546619137796e-05, "epoch": 1.61, "step": 43370}
+{"loss": 0.4675, "learning_rate": 5.9039768296758375e-05, "epoch": 1.61, "step": 43380}
+{"loss": 0.5813, "learning_rate": 5.89840704021388e-05, "epoch": 1.61, "step": 43390}
+{"loss": 0.5771, "learning_rate": 5.892837250751921e-05, "epoch": 1.61, "step": 43400}
+{"loss": 0.7069, "learning_rate": 5.887267461289963e-05, "epoch": 1.61, "step": 43410}
+{"loss": 0.6672, "learning_rate": 5.881697671828004e-05, "epoch": 1.61, "step": 43420}
+{"loss": 0.5319, "learning_rate": 5.876127882366046e-05, "epoch": 1.61, "step": 43430}
+{"loss": 0.622, "learning_rate": 5.870558092904088e-05, "epoch": 1.61, "step": 43440}
+{"loss": 0.6422, "learning_rate": 5.8649883034421294e-05, "epoch": 1.61, "step": 43450}
+{"loss": 0.6164, "learning_rate": 5.859418513980171e-05, "epoch": 1.61, "step": 43460}
+{"loss": 0.5758, "learning_rate": 5.853848724518213e-05, "epoch": 1.61, "step": 43470}
+{"loss": 0.5833, "learning_rate": 5.848278935056254e-05, "epoch": 1.61, "step": 43480}
+{"loss": 0.5216, "learning_rate": 5.8427091455942965e-05, "epoch": 1.61, "step": 43490}
+{"loss": 0.7481, "learning_rate": 5.8371393561323377e-05, "epoch": 1.61, "step": 43500}
+{"loss": 0.6219, "learning_rate": 5.8315695666703794e-05, "epoch": 1.61, "step": 43510}
+{"loss": 0.6885, "learning_rate": 5.825999777208421e-05, "epoch": 1.61, "step": 43520}
+{"loss": 0.6049, "learning_rate": 5.8204299877464624e-05, "epoch": 1.61, "step": 43530}
+{"loss": 0.5224, "learning_rate": 5.814860198284504e-05, "epoch": 1.61, "step": 43540}
+{"loss": 0.5185, "learning_rate": 5.809290408822546e-05, "epoch": 1.61, "step": 43550}
+{"loss": 0.685, "learning_rate": 5.803720619360588e-05, "epoch": 1.61, "step": 43560}
+{"loss": 0.6797, "learning_rate": 5.7981508298986295e-05, "epoch": 1.61, "step": 43570}
+{"loss": 0.5858, "learning_rate": 5.7925810404366706e-05, "epoch": 1.62, "step": 43580}
+{"loss": 0.5643, "learning_rate": 5.7870112509747124e-05, "epoch": 1.62, "step": 43590}
+{"loss": 0.5286, "learning_rate": 5.781441461512755e-05, "epoch": 1.62, "step": 43600}
+{"loss": 0.478, "learning_rate": 5.775871672050796e-05, "epoch": 1.62, "step": 43610}
+{"loss": 0.5956, "learning_rate": 5.770301882588838e-05, "epoch": 1.62, "step": 43620}
+{"loss": 0.5075, "learning_rate": 5.764732093126879e-05, "epoch": 1.62, "step": 43630}
+{"loss": 0.5067, "learning_rate": 5.759162303664921e-05, "epoch": 1.62, "step": 43640}
+{"loss": 0.6501, "learning_rate": 5.753592514202963e-05, "epoch": 1.62, "step": 43650}
+{"loss": 0.6296, "learning_rate": 5.748022724741004e-05, "epoch": 1.62, "step": 43660}
+{"loss": 0.5735, "learning_rate": 5.742452935279046e-05, "epoch": 1.62, "step": 43670}
+{"loss": 0.7211, "learning_rate": 5.736883145817088e-05, "epoch": 1.62, "step": 43680}
+{"loss": 0.6313, "learning_rate": 5.731313356355129e-05, "epoch": 1.62, "step": 43690}
+{"loss": 0.5505, "learning_rate": 5.7257435668931714e-05, "epoch": 1.62, "step": 43700}
+{"loss": 0.5815, "learning_rate": 5.7201737774312126e-05, "epoch": 1.62, "step": 43710}
+{"loss": 0.5547, "learning_rate": 5.7146039879692544e-05, "epoch": 1.62, "step": 43720}
+{"loss": 0.5678, "learning_rate": 5.709034198507296e-05, "epoch": 1.62, "step": 43730}
+{"loss": 0.5824, "learning_rate": 5.703464409045337e-05, "epoch": 1.62, "step": 43740}
+{"loss": 0.6383, "learning_rate": 5.69789461958338e-05, "epoch": 1.62, "step": 43750}
+{"loss": 0.5237, "learning_rate": 5.692324830121421e-05, "epoch": 1.62, "step": 43760}
+{"loss": 0.491, "learning_rate": 5.6867550406594626e-05, "epoch": 1.62, "step": 43770}
+{"loss": 0.5946, "learning_rate": 5.6811852511975044e-05, "epoch": 1.62, "step": 43780}
+{"loss": 0.5582, "learning_rate": 5.6756154617355455e-05, "epoch": 1.62, "step": 43790}
+{"loss": 0.6475, "learning_rate": 5.670045672273587e-05, "epoch": 1.62, "step": 43800}
+{"loss": 0.5595, "learning_rate": 5.66447588281163e-05, "epoch": 1.62, "step": 43810}
+{"loss": 0.5113, "learning_rate": 5.658906093349671e-05, "epoch": 1.62, "step": 43820}
+{"loss": 0.545, "learning_rate": 5.653336303887713e-05, "epoch": 1.62, "step": 43830}
+{"loss": 0.653, "learning_rate": 5.647766514425754e-05, "epoch": 1.62, "step": 43840}
+{"loss": 0.6246, "learning_rate": 5.6421967249637956e-05, "epoch": 1.63, "step": 43850}
+{"loss": 0.4969, "learning_rate": 5.636626935501838e-05, "epoch": 1.63, "step": 43860}
+{"loss": 0.6338, "learning_rate": 5.631057146039879e-05, "epoch": 1.63, "step": 43870}
+{"loss": 0.5956, "learning_rate": 5.625487356577921e-05, "epoch": 1.63, "step": 43880}
+{"loss": 0.522, "learning_rate": 5.619917567115962e-05, "epoch": 1.63, "step": 43890}
+{"loss": 0.7748, "learning_rate": 5.614347777654004e-05, "epoch": 1.63, "step": 43900}
+{"loss": 0.4356, "learning_rate": 5.6087779881920464e-05, "epoch": 1.63, "step": 43910}
+{"loss": 0.3851, "learning_rate": 5.6032081987300875e-05, "epoch": 1.63, "step": 43920}
+{"loss": 0.5757, "learning_rate": 5.597638409268129e-05, "epoch": 1.63, "step": 43930}
+{"loss": 0.441, "learning_rate": 5.592068619806171e-05, "epoch": 1.63, "step": 43940}
+{"loss": 0.548, "learning_rate": 5.586498830344212e-05, "epoch": 1.63, "step": 43950}
+{"loss": 0.4554, "learning_rate": 5.5809290408822546e-05, "epoch": 1.63, "step": 43960}
+{"loss": 0.4906, "learning_rate": 5.575359251420296e-05, "epoch": 1.63, "step": 43970}
+{"loss": 0.5354, "learning_rate": 5.5697894619583375e-05, "epoch": 1.63, "step": 43980}
+{"loss": 0.5002, "learning_rate": 5.564219672496379e-05, "epoch": 1.63, "step": 43990}
+{"loss": 0.5484, "learning_rate": 5.5586498830344204e-05, "epoch": 1.63, "step": 44000}
+{"loss": 0.5703, "learning_rate": 5.553080093572463e-05, "epoch": 1.63, "step": 44010}
+{"loss": 0.6664, "learning_rate": 5.547510304110504e-05, "epoch": 1.63, "step": 44020}
+{"loss": 0.4722, "learning_rate": 5.541940514648546e-05, "epoch": 1.63, "step": 44030}
+{"loss": 0.4626, "learning_rate": 5.5363707251865876e-05, "epoch": 1.63, "step": 44040}
+{"loss": 0.5479, "learning_rate": 5.530800935724629e-05, "epoch": 1.63, "step": 44050}
+{"loss": 0.525, "learning_rate": 5.5252311462626705e-05, "epoch": 1.63, "step": 44060}
+{"loss": 0.6428, "learning_rate": 5.519661356800713e-05, "epoch": 1.63, "step": 44070}
+{"loss": 0.5166, "learning_rate": 5.514091567338754e-05, "epoch": 1.63, "step": 44080}
+{"loss": 0.6326, "learning_rate": 5.508521777876796e-05, "epoch": 1.63, "step": 44090}
+{"loss": 0.5729, "learning_rate": 5.502951988414837e-05, "epoch": 1.63, "step": 44100}
+{"loss": 0.6143, "learning_rate": 5.497382198952879e-05, "epoch": 1.63, "step": 44110}
+{"loss": 0.6401, "learning_rate": 5.491812409490921e-05, "epoch": 1.64, "step": 44120}
+{"loss": 0.4702, "learning_rate": 5.4862426200289624e-05, "epoch": 1.64, "step": 44130}
+{"loss": 0.6347, "learning_rate": 5.480672830567004e-05, "epoch": 1.64, "step": 44140}
+{"loss": 0.5589, "learning_rate": 5.475103041105046e-05, "epoch": 1.64, "step": 44150}
+{"loss": 0.561, "learning_rate": 5.469533251643087e-05, "epoch": 1.64, "step": 44160}
+{"loss": 0.5932, "learning_rate": 5.4639634621811295e-05, "epoch": 1.64, "step": 44170}
+{"loss": 0.5326, "learning_rate": 5.4583936727191706e-05, "epoch": 1.64, "step": 44180}
+{"loss": 0.6079, "learning_rate": 5.4528238832572124e-05, "epoch": 1.64, "step": 44190}
+{"loss": 0.6451, "learning_rate": 5.447254093795254e-05, "epoch": 1.64, "step": 44200}
+{"loss": 0.7125, "learning_rate": 5.4416843043332953e-05, "epoch": 1.64, "step": 44210}
+{"loss": 0.5346, "learning_rate": 5.436114514871338e-05, "epoch": 1.64, "step": 44220}
+{"loss": 0.5436, "learning_rate": 5.430544725409379e-05, "epoch": 1.64, "step": 44230}
+{"loss": 0.7746, "learning_rate": 5.424974935947421e-05, "epoch": 1.64, "step": 44240}
+{"loss": 0.6162, "learning_rate": 5.4194051464854625e-05, "epoch": 1.64, "step": 44250}
+{"loss": 0.5579, "learning_rate": 5.4138353570235036e-05, "epoch": 1.64, "step": 44260}
+{"loss": 0.5597, "learning_rate": 5.408265567561546e-05, "epoch": 1.64, "step": 44270}
+{"loss": 0.594, "learning_rate": 5.402695778099588e-05, "epoch": 1.64, "step": 44280}
+{"loss": 0.5743, "learning_rate": 5.397125988637629e-05, "epoch": 1.64, "step": 44290}
+{"loss": 0.6061, "learning_rate": 5.391556199175671e-05, "epoch": 1.64, "step": 44300}
+{"loss": 0.5044, "learning_rate": 5.385986409713712e-05, "epoch": 1.64, "step": 44310}
+{"loss": 0.4822, "learning_rate": 5.380416620251754e-05, "epoch": 1.64, "step": 44320}
+{"loss": 0.5765, "learning_rate": 5.374846830789796e-05, "epoch": 1.64, "step": 44330}
+{"loss": 0.505, "learning_rate": 5.369277041327837e-05, "epoch": 1.64, "step": 44340}
+{"loss": 0.6134, "learning_rate": 5.363707251865879e-05, "epoch": 1.64, "step": 44350}
+{"loss": 0.5075, "learning_rate": 5.35813746240392e-05, "epoch": 1.64, "step": 44360}
+{"loss": 0.5673, "learning_rate": 5.352567672941962e-05, "epoch": 1.64, "step": 44370}
+{"loss": 0.6172, "learning_rate": 5.3469978834800044e-05, "epoch": 1.64, "step": 44380}
+{"loss": 0.5002, "learning_rate": 5.3414280940180456e-05, "epoch": 1.65, "step": 44390}
+{"loss": 0.6122, "learning_rate": 5.3358583045560873e-05, "epoch": 1.65, "step": 44400}
+{"loss": 0.5741, "learning_rate": 5.330288515094129e-05, "epoch": 1.65, "step": 44410}
+{"loss": 0.5744, "learning_rate": 5.32471872563217e-05, "epoch": 1.65, "step": 44420}
+{"loss": 0.5806, "learning_rate": 5.319148936170213e-05, "epoch": 1.65, "step": 44430}
+{"loss": 0.464, "learning_rate": 5.313579146708254e-05, "epoch": 1.65, "step": 44440}
+{"loss": 0.5507, "learning_rate": 5.3080093572462956e-05, "epoch": 1.65, "step": 44450}
+{"loss": 0.5791, "learning_rate": 5.302996546730533e-05, "epoch": 1.65, "step": 44460}
+{"loss": 0.592, "learning_rate": 5.297426757268575e-05, "epoch": 1.65, "step": 44470}
+{"loss": 0.636, "learning_rate": 5.291856967806616e-05, "epoch": 1.65, "step": 44480}
+{"loss": 0.6583, "learning_rate": 5.286287178344658e-05, "epoch": 1.65, "step": 44490}
+{"loss": 0.5768, "learning_rate": 5.2807173888827e-05, "epoch": 1.65, "step": 44500}
+{"loss": 0.5204, "learning_rate": 5.275147599420741e-05, "epoch": 1.65, "step": 44510}
+{"loss": 0.5837, "learning_rate": 5.269577809958783e-05, "epoch": 1.65, "step": 44520}
+{"loss": 0.5983, "learning_rate": 5.264008020496824e-05, "epoch": 1.65, "step": 44530}
+{"loss": 0.5578, "learning_rate": 5.2584382310348665e-05, "epoch": 1.65, "step": 44540}
+{"loss": 0.5907, "learning_rate": 5.2528684415729083e-05, "epoch": 1.65, "step": 44550}
+{"loss": 0.7457, "learning_rate": 5.2472986521109495e-05, "epoch": 1.65, "step": 44560}
+{"loss": 0.5937, "learning_rate": 5.241728862648991e-05, "epoch": 1.65, "step": 44570}
+{"loss": 0.6664, "learning_rate": 5.236159073187034e-05, "epoch": 1.65, "step": 44580}
+{"loss": 0.5235, "learning_rate": 5.230589283725075e-05, "epoch": 1.65, "step": 44590}
+{"loss": 0.7534, "learning_rate": 5.2250194942631166e-05, "epoch": 1.65, "step": 44600}
+{"loss": 0.6803, "learning_rate": 5.219449704801158e-05, "epoch": 1.65, "step": 44610}
+{"loss": 0.5623, "learning_rate": 5.2138799153391995e-05, "epoch": 1.65, "step": 44620}
+{"loss": 0.5566, "learning_rate": 5.208310125877242e-05, "epoch": 1.65, "step": 44630}
+{"loss": 0.6319, "learning_rate": 5.202740336415283e-05, "epoch": 1.65, "step": 44640}
+{"loss": 0.6485, "learning_rate": 5.197170546953325e-05, "epoch": 1.65, "step": 44650}
+{"loss": 0.5822, "learning_rate": 5.191600757491367e-05, "epoch": 1.66, "step": 44660}
+{"loss": 0.5371, "learning_rate": 5.186030968029408e-05, "epoch": 1.66, "step": 44670}
+{"loss": 0.4855, "learning_rate": 5.1804611785674496e-05, "epoch": 1.66, "step": 44680}
+{"loss": 0.538, "learning_rate": 5.1748913891054914e-05, "epoch": 1.66, "step": 44690}
+{"loss": 0.7173, "learning_rate": 5.169321599643533e-05, "epoch": 1.66, "step": 44700}
+{"loss": 0.6047, "learning_rate": 5.163751810181575e-05, "epoch": 1.66, "step": 44710}
+{"loss": 0.5279, "learning_rate": 5.158182020719616e-05, "epoch": 1.66, "step": 44720}
+{"loss": 0.5356, "learning_rate": 5.152612231257658e-05, "epoch": 1.66, "step": 44730}
+{"loss": 0.7094, "learning_rate": 5.147042441795699e-05, "epoch": 1.66, "step": 44740}
+{"loss": 0.4974, "learning_rate": 5.1414726523337415e-05, "epoch": 1.66, "step": 44750}
+{"loss": 0.5157, "learning_rate": 5.135902862871783e-05, "epoch": 1.66, "step": 44760}
+{"loss": 0.5486, "learning_rate": 5.1303330734098244e-05, "epoch": 1.66, "step": 44770}
+{"loss": 0.6363, "learning_rate": 5.124763283947866e-05, "epoch": 1.66, "step": 44780}
+{"loss": 0.4916, "learning_rate": 5.1191934944859086e-05, "epoch": 1.66, "step": 44790}
+{"loss": 0.5139, "learning_rate": 5.11362370502395e-05, "epoch": 1.66, "step": 44800}
+{"loss": 0.4681, "learning_rate": 5.1080539155619915e-05, "epoch": 1.66, "step": 44810}
+{"loss": 0.569, "learning_rate": 5.1024841261000326e-05, "epoch": 1.66, "step": 44820}
+{"loss": 0.5176, "learning_rate": 5.0969143366380744e-05, "epoch": 1.66, "step": 44830}
+{"loss": 0.6875, "learning_rate": 5.091344547176117e-05, "epoch": 1.66, "step": 44840}
+{"loss": 0.5992, "learning_rate": 5.085774757714158e-05, "epoch": 1.66, "step": 44850}
+{"loss": 0.6228, "learning_rate": 5.0802049682522e-05, "epoch": 1.66, "step": 44860}
+{"loss": 0.5097, "learning_rate": 5.074635178790241e-05, "epoch": 1.66, "step": 44870}
+{"loss": 0.5447, "learning_rate": 5.069065389328283e-05, "epoch": 1.66, "step": 44880}
+{"loss": 0.5607, "learning_rate": 5.063495599866325e-05, "epoch": 1.66, "step": 44890}
+{"loss": 0.5853, "learning_rate": 5.057925810404366e-05, "epoch": 1.66, "step": 44900}
+{"loss": 0.462, "learning_rate": 5.052356020942408e-05, "epoch": 1.66, "step": 44910}
+{"loss": 0.5318, "learning_rate": 5.04678623148045e-05, "epoch": 1.66, "step": 44920}
+{"loss": 0.538, "learning_rate": 5.041216442018491e-05, "epoch": 1.67, "step": 44930}
+{"loss": 0.53, "learning_rate": 5.035646652556533e-05, "epoch": 1.67, "step": 44940}
+{"loss": 0.5686, "learning_rate": 5.0300768630945746e-05, "epoch": 1.67, "step": 44950}
+{"loss": 0.5887, "learning_rate": 5.0245070736326164e-05, "epoch": 1.67, "step": 44960}
+{"loss": 0.5257, "learning_rate": 5.018937284170658e-05, "epoch": 1.67, "step": 44970}
+{"loss": 0.6711, "learning_rate": 5.013367494708699e-05, "epoch": 1.67, "step": 44980}
+{"loss": 0.538, "learning_rate": 5.007797705246741e-05, "epoch": 1.67, "step": 44990}
+{"loss": 0.5684, "learning_rate": 5.002227915784782e-05, "epoch": 1.67, "step": 45000}
+{"loss": 0.5431, "learning_rate": 4.9966581263228246e-05, "epoch": 1.67, "step": 45010}
+{"loss": 0.5484, "learning_rate": 4.9910883368608664e-05, "epoch": 1.67, "step": 45020}
+{"loss": 0.4859, "learning_rate": 4.9855185473989075e-05, "epoch": 1.67, "step": 45030}
+{"loss": 0.5966, "learning_rate": 4.979948757936949e-05, "epoch": 1.67, "step": 45040}
+{"loss": 0.5927, "learning_rate": 4.974378968474992e-05, "epoch": 1.67, "step": 45050}
+{"loss": 0.4849, "learning_rate": 4.968809179013033e-05, "epoch": 1.67, "step": 45060}
+{"loss": 0.5171, "learning_rate": 4.963239389551075e-05, "epoch": 1.67, "step": 45070}
+{"loss": 0.664, "learning_rate": 4.957669600089116e-05, "epoch": 1.67, "step": 45080}
+{"loss": 0.6009, "learning_rate": 4.9520998106271576e-05, "epoch": 1.67, "step": 45090}
+{"loss": 0.5988, "learning_rate": 4.9465300211652e-05, "epoch": 1.67, "step": 45100}
+{"loss": 0.5559, "learning_rate": 4.940960231703241e-05, "epoch": 1.67, "step": 45110}
+{"loss": 0.508, "learning_rate": 4.935390442241283e-05, "epoch": 1.67, "step": 45120}
+{"loss": 0.6703, "learning_rate": 4.929820652779325e-05, "epoch": 1.67, "step": 45130}
+{"loss": 0.5753, "learning_rate": 4.924250863317366e-05, "epoch": 1.67, "step": 45140}
+{"loss": 0.5609, "learning_rate": 4.9186810738554084e-05, "epoch": 1.67, "step": 45150}
+{"loss": 0.6179, "learning_rate": 4.9131112843934495e-05, "epoch": 1.67, "step": 45160}
+{"loss": 0.4829, "learning_rate": 4.907541494931491e-05, "epoch": 1.67, "step": 45170}
+{"loss": 0.5374, "learning_rate": 4.901971705469533e-05, "epoch": 1.67, "step": 45180}
+{"loss": 0.4785, "learning_rate": 4.896401916007574e-05, "epoch": 1.67, "step": 45190}
+{"loss": 0.4718, "learning_rate": 4.890832126545616e-05, "epoch": 1.68, "step": 45200}
+{"loss": 0.5101, "learning_rate": 4.885262337083658e-05, "epoch": 1.68, "step": 45210}
+{"loss": 0.6352, "learning_rate": 4.8796925476216995e-05, "epoch": 1.68, "step": 45220}
+{"loss": 0.4965, "learning_rate": 4.874122758159741e-05, "epoch": 1.68, "step": 45230}
+{"loss": 0.5496, "learning_rate": 4.8685529686977824e-05, "epoch": 1.68, "step": 45240}
+{"loss": 0.6111, "learning_rate": 4.862983179235824e-05, "epoch": 1.68, "step": 45250}
+{"loss": 0.6453, "learning_rate": 4.857413389773867e-05, "epoch": 1.68, "step": 45260}
+{"loss": 0.5842, "learning_rate": 4.851843600311908e-05, "epoch": 1.68, "step": 45270}
+{"loss": 0.5155, "learning_rate": 4.8462738108499496e-05, "epoch": 1.68, "step": 45280}
+{"loss": 0.6652, "learning_rate": 4.840704021387991e-05, "epoch": 1.68, "step": 45290}
+{"loss": 0.5439, "learning_rate": 4.8351342319260325e-05, "epoch": 1.68, "step": 45300}
+{"loss": 0.5285, "learning_rate": 4.829564442464075e-05, "epoch": 1.68, "step": 45310}
+{"loss": 0.605, "learning_rate": 4.823994653002116e-05, "epoch": 1.68, "step": 45320}
+{"loss": 0.5689, "learning_rate": 4.818424863540158e-05, "epoch": 1.68, "step": 45330}
+{"loss": 0.6165, "learning_rate": 4.812855074078199e-05, "epoch": 1.68, "step": 45340}
+{"loss": 0.5203, "learning_rate": 4.807285284616241e-05, "epoch": 1.68, "step": 45350}
+{"loss": 0.5506, "learning_rate": 4.801715495154283e-05, "epoch": 1.68, "step": 45360}
+{"loss": 0.4898, "learning_rate": 4.7961457056923244e-05, "epoch": 1.68, "step": 45370}
+{"loss": 0.5658, "learning_rate": 4.790575916230366e-05, "epoch": 1.68, "step": 45380}
+{"loss": 0.4978, "learning_rate": 4.785006126768408e-05, "epoch": 1.68, "step": 45390}
+{"loss": 0.5876, "learning_rate": 4.779436337306449e-05, "epoch": 1.68, "step": 45400}
+{"loss": 0.5455, "learning_rate": 4.7738665478444915e-05, "epoch": 1.68, "step": 45410}
+{"loss": 0.5432, "learning_rate": 4.7682967583825327e-05, "epoch": 1.68, "step": 45420}
+{"loss": 0.5204, "learning_rate": 4.7627269689205744e-05, "epoch": 1.68, "step": 45430}
+{"loss": 0.578, "learning_rate": 4.757157179458616e-05, "epoch": 1.68, "step": 45440}
+{"loss": 0.4366, "learning_rate": 4.7515873899966574e-05, "epoch": 1.68, "step": 45450}
+{"loss": 0.5416, "learning_rate": 4.7460176005347e-05, "epoch": 1.68, "step": 45460}
+{"loss": 0.5189, "learning_rate": 4.740447811072741e-05, "epoch": 1.69, "step": 45470}
+{"loss": 0.641, "learning_rate": 4.734878021610783e-05, "epoch": 1.69, "step": 45480}
+{"loss": 0.5771, "learning_rate": 4.7293082321488245e-05, "epoch": 1.69, "step": 45490}
+{"loss": 0.6699, "learning_rate": 4.7237384426868656e-05, "epoch": 1.69, "step": 45500}
+{"loss": 0.4789, "learning_rate": 4.7181686532249074e-05, "epoch": 1.69, "step": 45510}
+{"loss": 0.6007, "learning_rate": 4.71259886376295e-05, "epoch": 1.69, "step": 45520}
+{"loss": 0.608, "learning_rate": 4.707029074300991e-05, "epoch": 1.69, "step": 45530}
+{"loss": 0.4538, "learning_rate": 4.701459284839033e-05, "epoch": 1.69, "step": 45540}
+{"loss": 0.4578, "learning_rate": 4.695889495377074e-05, "epoch": 1.69, "step": 45550}
+{"loss": 0.5789, "learning_rate": 4.690319705915116e-05, "epoch": 1.69, "step": 45560}
+{"loss": 0.5112, "learning_rate": 4.684749916453158e-05, "epoch": 1.69, "step": 45570}
+{"loss": 0.5292, "learning_rate": 4.679180126991199e-05, "epoch": 1.69, "step": 45580}
+{"loss": 0.552, "learning_rate": 4.673610337529241e-05, "epoch": 1.69, "step": 45590}
+{"loss": 0.588, "learning_rate": 4.668040548067283e-05, "epoch": 1.69, "step": 45600}
+{"loss": 0.5784, "learning_rate": 4.662470758605324e-05, "epoch": 1.69, "step": 45610}
+{"loss": 0.7478, "learning_rate": 4.6569009691433664e-05, "epoch": 1.69, "step": 45620}
+{"loss": 0.5005, "learning_rate": 4.6513311796814076e-05, "epoch": 1.69, "step": 45630}
+{"loss": 0.6307, "learning_rate": 4.6457613902194494e-05, "epoch": 1.69, "step": 45640}
+{"loss": 0.486, "learning_rate": 4.640191600757491e-05, "epoch": 1.69, "step": 45650}
+{"loss": 0.5088, "learning_rate": 4.634621811295532e-05, "epoch": 1.69, "step": 45660}
+{"loss": 0.5886, "learning_rate": 4.629052021833575e-05, "epoch": 1.69, "step": 45670}
+{"loss": 0.4676, "learning_rate": 4.623482232371616e-05, "epoch": 1.69, "step": 45680}
+{"loss": 0.4557, "learning_rate": 4.6179124429096576e-05, "epoch": 1.69, "step": 45690}
+{"loss": 0.5614, "learning_rate": 4.6123426534476994e-05, "epoch": 1.69, "step": 45700}
+{"loss": 0.5603, "learning_rate": 4.6067728639857405e-05, "epoch": 1.69, "step": 45710}
+{"loss": 0.6486, "learning_rate": 4.601203074523783e-05, "epoch": 1.69, "step": 45720}
+{"loss": 0.6395, "learning_rate": 4.595633285061825e-05, "epoch": 1.69, "step": 45730}
+{"loss": 0.6031, "learning_rate": 4.590063495599866e-05, "epoch": 1.7, "step": 45740}
+{"loss": 0.6164, "learning_rate": 4.584493706137908e-05, "epoch": 1.7, "step": 45750}
+{"loss": 0.6158, "learning_rate": 4.578923916675949e-05, "epoch": 1.7, "step": 45760}
+{"loss": 0.4844, "learning_rate": 4.5733541272139906e-05, "epoch": 1.7, "step": 45770}
+{"loss": 0.5709, "learning_rate": 4.567784337752033e-05, "epoch": 1.7, "step": 45780}
+{"loss": 0.6556, "learning_rate": 4.562214548290074e-05, "epoch": 1.7, "step": 45790}
+{"loss": 0.495, "learning_rate": 4.556644758828116e-05, "epoch": 1.7, "step": 45800}
+{"loss": 0.5378, "learning_rate": 4.551074969366157e-05, "epoch": 1.7, "step": 45810}
+{"loss": 0.5235, "learning_rate": 4.545505179904199e-05, "epoch": 1.7, "step": 45820}
+{"loss": 0.61, "learning_rate": 4.5399353904422414e-05, "epoch": 1.7, "step": 45830}
+{"loss": 0.5406, "learning_rate": 4.5343656009802825e-05, "epoch": 1.7, "step": 45840}
+{"loss": 0.5562, "learning_rate": 4.528795811518324e-05, "epoch": 1.7, "step": 45850}
+{"loss": 0.4969, "learning_rate": 4.523226022056366e-05, "epoch": 1.7, "step": 45860}
+{"loss": 0.523, "learning_rate": 4.517656232594407e-05, "epoch": 1.7, "step": 45870}
+{"loss": 0.6725, "learning_rate": 4.5120864431324496e-05, "epoch": 1.7, "step": 45880}
+{"loss": 0.5714, "learning_rate": 4.506516653670491e-05, "epoch": 1.7, "step": 45890}
+{"loss": 0.5511, "learning_rate": 4.5009468642085325e-05, "epoch": 1.7, "step": 45900}
+{"loss": 0.5412, "learning_rate": 4.495377074746574e-05, "epoch": 1.7, "step": 45910}
+{"loss": 0.4855, "learning_rate": 4.4898072852846154e-05, "epoch": 1.7, "step": 45920}
+{"loss": 0.5127, "learning_rate": 4.484237495822658e-05, "epoch": 1.7, "step": 45930}
+{"loss": 0.6152, "learning_rate": 4.478667706360699e-05, "epoch": 1.7, "step": 45940}
+{"loss": 0.6062, "learning_rate": 4.473097916898741e-05, "epoch": 1.7, "step": 45950}
+{"loss": 0.5077, "learning_rate": 4.4675281274367826e-05, "epoch": 1.7, "step": 45960}
+{"loss": 0.5717, "learning_rate": 4.461958337974824e-05, "epoch": 1.7, "step": 45970}
+{"loss": 0.5763, "learning_rate": 4.456388548512866e-05, "epoch": 1.7, "step": 45980}
+{"loss": 0.6363, "learning_rate": 4.450818759050908e-05, "epoch": 1.7, "step": 45990}
+{"loss": 0.545, "learning_rate": 4.445248969588949e-05, "epoch": 1.7, "step": 46000}
+{"loss": 0.5906, "learning_rate": 4.439679180126991e-05, "epoch": 1.71, "step": 46010}
+{"loss": 0.6867, "learning_rate": 4.434109390665032e-05, "epoch": 1.71, "step": 46020}
+{"loss": 0.6252, "learning_rate": 4.428539601203074e-05, "epoch": 1.71, "step": 46030}
+{"loss": 0.4837, "learning_rate": 4.422969811741116e-05, "epoch": 1.71, "step": 46040}
+{"loss": 0.4988, "learning_rate": 4.4174000222791574e-05, "epoch": 1.71, "step": 46050}
+{"loss": 0.5887, "learning_rate": 4.411830232817199e-05, "epoch": 1.71, "step": 46060}
+{"loss": 0.5581, "learning_rate": 4.406260443355241e-05, "epoch": 1.71, "step": 46070}
+{"loss": 0.6246, "learning_rate": 4.400690653893282e-05, "epoch": 1.71, "step": 46080}
+{"loss": 0.5884, "learning_rate": 4.3951208644313245e-05, "epoch": 1.71, "step": 46090}
+{"loss": 0.5568, "learning_rate": 4.3895510749693656e-05, "epoch": 1.71, "step": 46100}
+{"loss": 0.6271, "learning_rate": 4.3839812855074074e-05, "epoch": 1.71, "step": 46110}
+{"loss": 0.6573, "learning_rate": 4.378411496045449e-05, "epoch": 1.71, "step": 46120}
+{"loss": 0.7016, "learning_rate": 4.3728417065834903e-05, "epoch": 1.71, "step": 46130}
+{"loss": 0.7095, "learning_rate": 4.367271917121533e-05, "epoch": 1.71, "step": 46140}
+{"loss": 0.4233, "learning_rate": 4.361702127659574e-05, "epoch": 1.71, "step": 46150}
+{"loss": 0.504, "learning_rate": 4.356132338197616e-05, "epoch": 1.71, "step": 46160}
+{"loss": 0.5638, "learning_rate": 4.3505625487356575e-05, "epoch": 1.71, "step": 46170}
+{"loss": 0.5206, "learning_rate": 4.3449927592736986e-05, "epoch": 1.71, "step": 46180}
+{"loss": 0.4868, "learning_rate": 4.339422969811741e-05, "epoch": 1.71, "step": 46190}
+{"loss": 0.4563, "learning_rate": 4.333853180349783e-05, "epoch": 1.71, "step": 46200}
+{"loss": 0.6311, "learning_rate": 4.328283390887824e-05, "epoch": 1.71, "step": 46210}
+{"loss": 0.5634, "learning_rate": 4.322713601425866e-05, "epoch": 1.71, "step": 46220}
+{"loss": 0.6034, "learning_rate": 4.317143811963907e-05, "epoch": 1.71, "step": 46230}
+{"loss": 0.6102, "learning_rate": 4.3115740225019494e-05, "epoch": 1.71, "step": 46240}
+{"loss": 0.6777, "learning_rate": 4.306004233039991e-05, "epoch": 1.71, "step": 46250}
+{"loss": 0.6366, "learning_rate": 4.300434443578032e-05, "epoch": 1.71, "step": 46260}
+{"loss": 0.7463, "learning_rate": 4.294864654116074e-05, "epoch": 1.71, "step": 46270}
+{"loss": 0.6662, "learning_rate": 4.289294864654115e-05, "epoch": 1.72, "step": 46280}
+{"loss": 0.7086, "learning_rate": 4.283725075192157e-05, "epoch": 1.72, "step": 46290}
+{"loss": 0.5628, "learning_rate": 4.2781552857301994e-05, "epoch": 1.72, "step": 46300}
+{"loss": 0.5446, "learning_rate": 4.2725854962682406e-05, "epoch": 1.72, "step": 46310}
+{"loss": 0.6231, "learning_rate": 4.2670157068062823e-05, "epoch": 1.72, "step": 46320}
+{"loss": 0.6283, "learning_rate": 4.261445917344324e-05, "epoch": 1.72, "step": 46330}
+{"loss": 0.4289, "learning_rate": 4.255876127882365e-05, "epoch": 1.72, "step": 46340}
+{"loss": 0.5489, "learning_rate": 4.250306338420408e-05, "epoch": 1.72, "step": 46350}
+{"loss": 0.5115, "learning_rate": 4.244736548958449e-05, "epoch": 1.72, "step": 46360}
+{"loss": 0.5882, "learning_rate": 4.2391667594964906e-05, "epoch": 1.72, "step": 46370}
+{"loss": 0.6087, "learning_rate": 4.2335969700345324e-05, "epoch": 1.72, "step": 46380}
+{"loss": 0.6942, "learning_rate": 4.2280271805725735e-05, "epoch": 1.72, "step": 46390}
+{"loss": 0.6756, "learning_rate": 4.222457391110616e-05, "epoch": 1.72, "step": 46400}
+{"loss": 0.6181, "learning_rate": 4.216887601648657e-05, "epoch": 1.72, "step": 46410}
+{"loss": 0.6081, "learning_rate": 4.211317812186699e-05, "epoch": 1.72, "step": 46420}
+{"loss": 0.5893, "learning_rate": 4.205748022724741e-05, "epoch": 1.72, "step": 46430}
+{"loss": 0.5367, "learning_rate": 4.200178233262782e-05, "epoch": 1.72, "step": 46440}
+{"loss": 0.5065, "learning_rate": 4.194608443800824e-05, "epoch": 1.72, "step": 46450}
+{"loss": 0.5657, "learning_rate": 4.189038654338866e-05, "epoch": 1.72, "step": 46460}
+{"loss": 0.5578, "learning_rate": 4.183468864876907e-05, "epoch": 1.72, "step": 46470}
+{"loss": 0.5015, "learning_rate": 4.177899075414949e-05, "epoch": 1.72, "step": 46480}
+{"loss": 0.6406, "learning_rate": 4.17232928595299e-05, "epoch": 1.72, "step": 46490}
+{"loss": 0.5087, "learning_rate": 4.1667594964910326e-05, "epoch": 1.72, "step": 46500}
+{"loss": 0.5553, "learning_rate": 4.1611897070290743e-05, "epoch": 1.72, "step": 46510}
+{"loss": 0.669, "learning_rate": 4.1556199175671155e-05, "epoch": 1.72, "step": 46520}
+{"loss": 0.4915, "learning_rate": 4.150050128105157e-05, "epoch": 1.72, "step": 46530}
+{"loss": 0.5879, "learning_rate": 4.144480338643199e-05, "epoch": 1.72, "step": 46540}
+{"loss": 0.664, "learning_rate": 4.139467528127437e-05, "epoch": 1.73, "step": 46550}
+{"loss": 0.5826, "learning_rate": 4.133897738665478e-05, "epoch": 1.73, "step": 46560}
+{"loss": 0.6061, "learning_rate": 4.12832794920352e-05, "epoch": 1.73, "step": 46570}
+{"loss": 0.4465, "learning_rate": 4.122758159741562e-05, "epoch": 1.73, "step": 46580}
+{"loss": 0.5513, "learning_rate": 4.117188370279603e-05, "epoch": 1.73, "step": 46590}
+{"loss": 0.4745, "learning_rate": 4.111618580817645e-05, "epoch": 1.73, "step": 46600}
+{"loss": 0.6092, "learning_rate": 4.1060487913556864e-05, "epoch": 1.73, "step": 46610}
+{"loss": 0.7792, "learning_rate": 4.100479001893728e-05, "epoch": 1.73, "step": 46620}
+{"loss": 0.6153, "learning_rate": 4.09490921243177e-05, "epoch": 1.73, "step": 46630}
+{"loss": 0.5581, "learning_rate": 4.089339422969811e-05, "epoch": 1.73, "step": 46640}
+{"loss": 0.526, "learning_rate": 4.083769633507853e-05, "epoch": 1.73, "step": 46650}
+{"loss": 0.5867, "learning_rate": 4.078199844045895e-05, "epoch": 1.73, "step": 46660}
+{"loss": 0.5199, "learning_rate": 4.0726300545839365e-05, "epoch": 1.73, "step": 46670}
+{"loss": 0.496, "learning_rate": 4.067060265121978e-05, "epoch": 1.73, "step": 46680}
+{"loss": 0.5653, "learning_rate": 4.0614904756600194e-05, "epoch": 1.73, "step": 46690}
+{"loss": 0.6168, "learning_rate": 4.055920686198061e-05, "epoch": 1.73, "step": 46700}
+{"loss": 0.5163, "learning_rate": 4.0503508967361036e-05, "epoch": 1.73, "step": 46710}
+{"loss": 0.5686, "learning_rate": 4.044781107274145e-05, "epoch": 1.73, "step": 46720}
+{"loss": 0.6113, "learning_rate": 4.0392113178121865e-05, "epoch": 1.73, "step": 46730}
+{"loss": 0.5519, "learning_rate": 4.0336415283502276e-05, "epoch": 1.73, "step": 46740}
+{"loss": 0.4987, "learning_rate": 4.0280717388882694e-05, "epoch": 1.73, "step": 46750}
+{"loss": 0.5284, "learning_rate": 4.022501949426312e-05, "epoch": 1.73, "step": 46760}
+{"loss": 0.6033, "learning_rate": 4.016932159964353e-05, "epoch": 1.73, "step": 46770}
+{"loss": 0.6149, "learning_rate": 4.011362370502395e-05, "epoch": 1.73, "step": 46780}
+{"loss": 0.5985, "learning_rate": 4.005792581040436e-05, "epoch": 1.73, "step": 46790}
+{"loss": 0.6426, "learning_rate": 4.000222791578478e-05, "epoch": 1.73, "step": 46800}
+{"loss": 0.6223, "learning_rate": 3.99465300211652e-05, "epoch": 1.73, "step": 46810}
+{"loss": 0.5255, "learning_rate": 3.989083212654561e-05, "epoch": 1.74, "step": 46820}
+{"loss": 0.564, "learning_rate": 3.983513423192603e-05, "epoch": 1.74, "step": 46830}
+{"loss": 0.5405, "learning_rate": 3.977943633730645e-05, "epoch": 1.74, "step": 46840}
+{"loss": 0.5923, "learning_rate": 3.972373844268686e-05, "epoch": 1.74, "step": 46850}
+{"loss": 0.5136, "learning_rate": 3.9668040548067285e-05, "epoch": 1.74, "step": 46860}
+{"loss": 0.5915, "learning_rate": 3.9612342653447696e-05, "epoch": 1.74, "step": 46870}
+{"loss": 0.4871, "learning_rate": 3.9556644758828114e-05, "epoch": 1.74, "step": 46880}
+{"loss": 0.5138, "learning_rate": 3.950094686420853e-05, "epoch": 1.74, "step": 46890}
+{"loss": 0.4976, "learning_rate": 3.944524896958894e-05, "epoch": 1.74, "step": 46900}
+{"loss": 0.4912, "learning_rate": 3.938955107496936e-05, "epoch": 1.74, "step": 46910}
+{"loss": 0.7713, "learning_rate": 3.933385318034978e-05, "epoch": 1.74, "step": 46920}
+{"loss": 0.6141, "learning_rate": 3.9278155285730196e-05, "epoch": 1.74, "step": 46930}
+{"loss": 0.5806, "learning_rate": 3.9222457391110614e-05, "epoch": 1.74, "step": 46940}
+{"loss": 0.5268, "learning_rate": 3.9166759496491025e-05, "epoch": 1.74, "step": 46950}
+{"loss": 0.6086, "learning_rate": 3.911106160187144e-05, "epoch": 1.74, "step": 46960}
+{"loss": 0.4905, "learning_rate": 3.905536370725187e-05, "epoch": 1.74, "step": 46970}
+{"loss": 0.6449, "learning_rate": 3.899966581263228e-05, "epoch": 1.74, "step": 46980}
+{"loss": 0.5924, "learning_rate": 3.89439679180127e-05, "epoch": 1.74, "step": 46990}
+{"loss": 0.6073, "learning_rate": 3.888827002339311e-05, "epoch": 1.74, "step": 47000}
+{"loss": 0.5415, "learning_rate": 3.8832572128773526e-05, "epoch": 1.74, "step": 47010}
+{"loss": 0.5702, "learning_rate": 3.877687423415395e-05, "epoch": 1.74, "step": 47020}
+{"loss": 0.4638, "learning_rate": 3.872117633953436e-05, "epoch": 1.74, "step": 47030}
+{"loss": 0.6281, "learning_rate": 3.866547844491478e-05, "epoch": 1.74, "step": 47040}
+{"loss": 0.5637, "learning_rate": 3.86097805502952e-05, "epoch": 1.74, "step": 47050}
+{"loss": 0.5895, "learning_rate": 3.855408265567561e-05, "epoch": 1.74, "step": 47060}
+{"loss": 0.3958, "learning_rate": 3.8498384761056034e-05, "epoch": 1.74, "step": 47070}
+{"loss": 0.4596, "learning_rate": 3.8442686866436445e-05, "epoch": 1.74, "step": 47080}
+{"loss": 0.5062, "learning_rate": 3.838698897181686e-05, "epoch": 1.75, "step": 47090}
+{"loss": 0.6428, "learning_rate": 3.833129107719728e-05, "epoch": 1.75, "step": 47100}
+{"loss": 0.5513, "learning_rate": 3.827559318257769e-05, "epoch": 1.75, "step": 47110}
+{"loss": 0.5119, "learning_rate": 3.8219895287958116e-05, "epoch": 1.75, "step": 47120}
+{"loss": 0.5603, "learning_rate": 3.816419739333853e-05, "epoch": 1.75, "step": 47130}
+{"loss": 0.6028, "learning_rate": 3.8108499498718945e-05, "epoch": 1.75, "step": 47140}
+{"loss": 0.6344, "learning_rate": 3.805280160409936e-05, "epoch": 1.75, "step": 47150}
+{"loss": 0.5946, "learning_rate": 3.7997103709479774e-05, "epoch": 1.75, "step": 47160}
+{"loss": 0.5356, "learning_rate": 3.794140581486019e-05, "epoch": 1.75, "step": 47170}
+{"loss": 0.5491, "learning_rate": 3.788570792024062e-05, "epoch": 1.75, "step": 47180}
+{"loss": 0.5036, "learning_rate": 3.783001002562103e-05, "epoch": 1.75, "step": 47190}
+{"loss": 0.5586, "learning_rate": 3.7774312131001446e-05, "epoch": 1.75, "step": 47200}
+{"loss": 0.5447, "learning_rate": 3.771861423638186e-05, "epoch": 1.75, "step": 47210}
+{"loss": 0.4533, "learning_rate": 3.7662916341762275e-05, "epoch": 1.75, "step": 47220}
+{"loss": 0.5152, "learning_rate": 3.76072184471427e-05, "epoch": 1.75, "step": 47230}
+{"loss": 0.5746, "learning_rate": 3.755152055252311e-05, "epoch": 1.75, "step": 47240}
+{"loss": 0.5726, "learning_rate": 3.749582265790353e-05, "epoch": 1.75, "step": 47250}
+{"loss": 0.6571, "learning_rate": 3.744012476328395e-05, "epoch": 1.75, "step": 47260}
+{"loss": 0.5132, "learning_rate": 3.738442686866436e-05, "epoch": 1.75, "step": 47270}
+{"loss": 0.5532, "learning_rate": 3.7328728974044776e-05, "epoch": 1.75, "step": 47280}
+{"loss": 0.5015, "learning_rate": 3.7273031079425194e-05, "epoch": 1.75, "step": 47290}
+{"loss": 0.4945, "learning_rate": 3.721733318480561e-05, "epoch": 1.75, "step": 47300}
+{"loss": 0.6086, "learning_rate": 3.716163529018603e-05, "epoch": 1.75, "step": 47310}
+{"loss": 0.4906, "learning_rate": 3.710593739556644e-05, "epoch": 1.75, "step": 47320}
+{"loss": 0.6731, "learning_rate": 3.7050239500946865e-05, "epoch": 1.75, "step": 47330}
+{"loss": 0.5482, "learning_rate": 3.6994541606327277e-05, "epoch": 1.75, "step": 47340}
+{"loss": 0.4853, "learning_rate": 3.6938843711707694e-05, "epoch": 1.75, "step": 47350}
+{"loss": 0.487, "learning_rate": 3.688314581708811e-05, "epoch": 1.76, "step": 47360}
+{"loss": 0.5661, "learning_rate": 3.682744792246853e-05, "epoch": 1.76, "step": 47370}
+{"loss": 0.4321, "learning_rate": 3.677175002784895e-05, "epoch": 1.76, "step": 47380}
+{"loss": 0.6578, "learning_rate": 3.671605213322936e-05, "epoch": 1.76, "step": 47390}
+{"loss": 0.654, "learning_rate": 3.666035423860978e-05, "epoch": 1.76, "step": 47400}
+{"loss": 0.5783, "learning_rate": 3.6604656343990195e-05, "epoch": 1.76, "step": 47410}
+{"loss": 0.4982, "learning_rate": 3.654895844937061e-05, "epoch": 1.76, "step": 47420}
+{"loss": 0.5617, "learning_rate": 3.649326055475103e-05, "epoch": 1.76, "step": 47430}
+{"loss": 0.5977, "learning_rate": 3.643756266013144e-05, "epoch": 1.76, "step": 47440}
+{"loss": 0.5348, "learning_rate": 3.638186476551186e-05, "epoch": 1.76, "step": 47450}
+{"loss": 0.5685, "learning_rate": 3.632616687089228e-05, "epoch": 1.76, "step": 47460}
+{"loss": 0.5385, "learning_rate": 3.6270468976272696e-05, "epoch": 1.76, "step": 47470}
+{"loss": 0.6513, "learning_rate": 3.621477108165311e-05, "epoch": 1.76, "step": 47480}
+{"loss": 0.5679, "learning_rate": 3.6159073187033525e-05, "epoch": 1.76, "step": 47490}
+{"loss": 0.502, "learning_rate": 3.610337529241394e-05, "epoch": 1.76, "step": 47500}
+{"loss": 0.5285, "learning_rate": 3.604767739779436e-05, "epoch": 1.76, "step": 47510}
+{"loss": 0.405, "learning_rate": 3.599197950317478e-05, "epoch": 1.76, "step": 47520}
+{"loss": 0.6419, "learning_rate": 3.593628160855519e-05, "epoch": 1.76, "step": 47530}
+{"loss": 0.5418, "learning_rate": 3.588058371393561e-05, "epoch": 1.76, "step": 47540}
+{"loss": 0.515, "learning_rate": 3.5824885819316026e-05, "epoch": 1.76, "step": 47550}
+{"loss": 0.4902, "learning_rate": 3.5769187924696444e-05, "epoch": 1.76, "step": 47560}
+{"loss": 0.5365, "learning_rate": 3.571349003007686e-05, "epoch": 1.76, "step": 47570}
+{"loss": 0.5141, "learning_rate": 3.565779213545727e-05, "epoch": 1.76, "step": 47580}
+{"loss": 0.6049, "learning_rate": 3.56020942408377e-05, "epoch": 1.76, "step": 47590}
+{"loss": 0.5635, "learning_rate": 3.554639634621811e-05, "epoch": 1.76, "step": 47600}
+{"loss": 0.6949, "learning_rate": 3.5490698451598526e-05, "epoch": 1.76, "step": 47610}
+{"loss": 0.5147, "learning_rate": 3.5435000556978944e-05, "epoch": 1.76, "step": 47620}
+{"loss": 0.6236, "learning_rate": 3.537930266235936e-05, "epoch": 1.77, "step": 47630}
+{"loss": 0.7037, "learning_rate": 3.532360476773978e-05, "epoch": 1.77, "step": 47640}
+{"loss": 0.5828, "learning_rate": 3.526790687312019e-05, "epoch": 1.77, "step": 47650}
+{"loss": 0.51, "learning_rate": 3.521220897850061e-05, "epoch": 1.77, "step": 47660}
+{"loss": 0.5796, "learning_rate": 3.515651108388103e-05, "epoch": 1.77, "step": 47670}
+{"loss": 0.5492, "learning_rate": 3.5100813189261445e-05, "epoch": 1.77, "step": 47680}
+{"loss": 0.6647, "learning_rate": 3.504511529464186e-05, "epoch": 1.77, "step": 47690}
+{"loss": 0.5784, "learning_rate": 3.4989417400022274e-05, "epoch": 1.77, "step": 47700}
+{"loss": 0.7018, "learning_rate": 3.493371950540269e-05, "epoch": 1.77, "step": 47710}
+{"loss": 0.6299, "learning_rate": 3.487802161078311e-05, "epoch": 1.77, "step": 47720}
+{"loss": 0.5654, "learning_rate": 3.482232371616353e-05, "epoch": 1.77, "step": 47730}
+{"loss": 0.5152, "learning_rate": 3.476662582154394e-05, "epoch": 1.77, "step": 47740}
+{"loss": 0.5748, "learning_rate": 3.471092792692436e-05, "epoch": 1.77, "step": 47750}
+{"loss": 0.5715, "learning_rate": 3.4655230032304775e-05, "epoch": 1.77, "step": 47760}
+{"loss": 0.4943, "learning_rate": 3.459953213768519e-05, "epoch": 1.77, "step": 47770}
+{"loss": 0.5994, "learning_rate": 3.454383424306561e-05, "epoch": 1.77, "step": 47780}
+{"loss": 0.614, "learning_rate": 3.448813634844602e-05, "epoch": 1.77, "step": 47790}
+{"loss": 0.6671, "learning_rate": 3.4432438453826446e-05, "epoch": 1.77, "step": 47800}
+{"loss": 0.5859, "learning_rate": 3.437674055920686e-05, "epoch": 1.77, "step": 47810}
+{"loss": 0.5051, "learning_rate": 3.4321042664587275e-05, "epoch": 1.77, "step": 47820}
+{"loss": 0.6863, "learning_rate": 3.426534476996769e-05, "epoch": 1.77, "step": 47830}
+{"loss": 0.5615, "learning_rate": 3.4209646875348104e-05, "epoch": 1.77, "step": 47840}
+{"loss": 0.558, "learning_rate": 3.415394898072853e-05, "epoch": 1.77, "step": 47850}
+{"loss": 0.5991, "learning_rate": 3.409825108610894e-05, "epoch": 1.77, "step": 47860}
+{"loss": 0.6507, "learning_rate": 3.404255319148936e-05, "epoch": 1.77, "step": 47870}
+{"loss": 0.5231, "learning_rate": 3.3986855296869776e-05, "epoch": 1.77, "step": 47880}
+{"loss": 0.6547, "learning_rate": 3.3931157402250194e-05, "epoch": 1.77, "step": 47890}
+{"loss": 0.6191, "learning_rate": 3.387545950763061e-05, "epoch": 1.78, "step": 47900}
+{"loss": 0.5995, "learning_rate": 3.381976161301102e-05, "epoch": 1.78, "step": 47910}
+{"loss": 0.6403, "learning_rate": 3.376406371839144e-05, "epoch": 1.78, "step": 47920}
+{"loss": 0.5227, "learning_rate": 3.370836582377186e-05, "epoch": 1.78, "step": 47930}
+{"loss": 0.6298, "learning_rate": 3.365266792915228e-05, "epoch": 1.78, "step": 47940}
+{"loss": 0.5616, "learning_rate": 3.3596970034532695e-05, "epoch": 1.78, "step": 47950}
+{"loss": 0.563, "learning_rate": 3.3541272139913106e-05, "epoch": 1.78, "step": 47960}
+{"loss": 0.6102, "learning_rate": 3.3485574245293524e-05, "epoch": 1.78, "step": 47970}
+{"loss": 0.5515, "learning_rate": 3.342987635067394e-05, "epoch": 1.78, "step": 47980}
+{"loss": 0.5193, "learning_rate": 3.337417845605436e-05, "epoch": 1.78, "step": 47990}
+{"loss": 0.6968, "learning_rate": 3.331848056143477e-05, "epoch": 1.78, "step": 48000}
+{"loss": 0.5517, "learning_rate": 3.326278266681519e-05, "epoch": 1.78, "step": 48010}
+{"loss": 0.6192, "learning_rate": 3.3207084772195606e-05, "epoch": 1.78, "step": 48020}
+{"loss": 0.6097, "learning_rate": 3.3151386877576024e-05, "epoch": 1.78, "step": 48030}
+{"loss": 0.5692, "learning_rate": 3.309568898295644e-05, "epoch": 1.78, "step": 48040}
+{"loss": 0.5861, "learning_rate": 3.3039991088336853e-05, "epoch": 1.78, "step": 48050}
+{"loss": 0.5805, "learning_rate": 3.298429319371728e-05, "epoch": 1.78, "step": 48060}
+{"loss": 0.5638, "learning_rate": 3.292859529909769e-05, "epoch": 1.78, "step": 48070}
+{"loss": 0.5188, "learning_rate": 3.287289740447811e-05, "epoch": 1.78, "step": 48080}
+{"loss": 0.5722, "learning_rate": 3.2817199509858525e-05, "epoch": 1.78, "step": 48090}
+{"loss": 0.5402, "learning_rate": 3.276150161523894e-05, "epoch": 1.78, "step": 48100}
+{"loss": 0.5593, "learning_rate": 3.270580372061936e-05, "epoch": 1.78, "step": 48110}
+{"loss": 0.7037, "learning_rate": 3.265010582599977e-05, "epoch": 1.78, "step": 48120}
+{"loss": 0.5058, "learning_rate": 3.259440793138019e-05, "epoch": 1.78, "step": 48130}
+{"loss": 0.5378, "learning_rate": 3.253871003676061e-05, "epoch": 1.78, "step": 48140}
+{"loss": 0.6221, "learning_rate": 3.2483012142141026e-05, "epoch": 1.78, "step": 48150}
+{"loss": 0.5499, "learning_rate": 3.2427314247521444e-05, "epoch": 1.78, "step": 48160}
+{"loss": 0.4452, "learning_rate": 3.2371616352901855e-05, "epoch": 1.79, "step": 48170}
+{"loss": 0.4904, "learning_rate": 3.231591845828227e-05, "epoch": 1.79, "step": 48180}
+{"loss": 0.6345, "learning_rate": 3.226022056366269e-05, "epoch": 1.79, "step": 48190}
+{"loss": 0.7162, "learning_rate": 3.220452266904311e-05, "epoch": 1.79, "step": 48200}
+{"loss": 0.4464, "learning_rate": 3.2148824774423526e-05, "epoch": 1.79, "step": 48210}
+{"loss": 0.5543, "learning_rate": 3.209312687980394e-05, "epoch": 1.79, "step": 48220}
+{"loss": 0.5664, "learning_rate": 3.2037428985184356e-05, "epoch": 1.79, "step": 48230}
+{"loss": 0.5523, "learning_rate": 3.1981731090564773e-05, "epoch": 1.79, "step": 48240}
+{"loss": 0.5612, "learning_rate": 3.192603319594519e-05, "epoch": 1.79, "step": 48250}
+{"loss": 0.4993, "learning_rate": 3.18703353013256e-05, "epoch": 1.79, "step": 48260}
+{"loss": 0.6767, "learning_rate": 3.181463740670603e-05, "epoch": 1.79, "step": 48270}
+{"loss": 0.5287, "learning_rate": 3.175893951208644e-05, "epoch": 1.79, "step": 48280}
+{"loss": 0.5508, "learning_rate": 3.1703241617466856e-05, "epoch": 1.79, "step": 48290}
+{"loss": 0.6466, "learning_rate": 3.1647543722847274e-05, "epoch": 1.79, "step": 48300}
+{"loss": 0.5184, "learning_rate": 3.1591845828227685e-05, "epoch": 1.79, "step": 48310}
+{"loss": 0.4867, "learning_rate": 3.153614793360811e-05, "epoch": 1.79, "step": 48320}
+{"loss": 0.5099, "learning_rate": 3.148045003898852e-05, "epoch": 1.79, "step": 48330}
+{"loss": 0.5857, "learning_rate": 3.142475214436894e-05, "epoch": 1.79, "step": 48340}
+{"loss": 0.5391, "learning_rate": 3.136905424974936e-05, "epoch": 1.79, "step": 48350}
+{"loss": 0.5064, "learning_rate": 3.1313356355129775e-05, "epoch": 1.79, "step": 48360}
+{"loss": 0.6499, "learning_rate": 3.125765846051019e-05, "epoch": 1.79, "step": 48370}
+{"loss": 0.5281, "learning_rate": 3.1201960565890604e-05, "epoch": 1.79, "step": 48380}
+{"loss": 0.6135, "learning_rate": 3.114626267127102e-05, "epoch": 1.79, "step": 48390}
+{"loss": 0.5571, "learning_rate": 3.109056477665144e-05, "epoch": 1.79, "step": 48400}
+{"loss": 0.6009, "learning_rate": 3.103486688203186e-05, "epoch": 1.79, "step": 48410}
+{"loss": 0.523, "learning_rate": 3.0979168987412276e-05, "epoch": 1.79, "step": 48420}
+{"loss": 0.4927, "learning_rate": 3.092347109279269e-05, "epoch": 1.79, "step": 48430}
+{"loss": 0.739, "learning_rate": 3.086777319817311e-05, "epoch": 1.8, "step": 48440}
+{"loss": 0.4902, "learning_rate": 3.081207530355352e-05, "epoch": 1.8, "step": 48450}
+{"loss": 0.5542, "learning_rate": 3.075637740893394e-05, "epoch": 1.8, "step": 48460}
+{"loss": 0.484, "learning_rate": 3.070067951431436e-05, "epoch": 1.8, "step": 48470}
+{"loss": 0.5727, "learning_rate": 3.064498161969477e-05, "epoch": 1.8, "step": 48480}
+{"loss": 0.5418, "learning_rate": 3.058928372507519e-05, "epoch": 1.8, "step": 48490}
+{"loss": 0.5902, "learning_rate": 3.0533585830455605e-05, "epoch": 1.8, "step": 48500}
+{"loss": 0.5889, "learning_rate": 3.0477887935836023e-05, "epoch": 1.8, "step": 48510}
+{"loss": 0.6031, "learning_rate": 3.0422190041216438e-05, "epoch": 1.8, "step": 48520}
+{"loss": 0.5224, "learning_rate": 3.036649214659686e-05, "epoch": 1.8, "step": 48530}
+{"loss": 0.5233, "learning_rate": 3.0310794251977274e-05, "epoch": 1.8, "step": 48540}
+{"loss": 0.542, "learning_rate": 3.0255096357357688e-05, "epoch": 1.8, "step": 48550}
+{"loss": 0.531, "learning_rate": 3.0199398462738106e-05, "epoch": 1.8, "step": 48560}
+{"loss": 0.6225, "learning_rate": 3.0149270357580482e-05, "epoch": 1.8, "step": 48570}
+{"loss": 0.6596, "learning_rate": 3.0093572462960897e-05, "epoch": 1.8, "step": 48580}
+{"loss": 0.7282, "learning_rate": 3.0037874568341315e-05, "epoch": 1.8, "step": 48590}
+{"loss": 0.5325, "learning_rate": 2.998217667372173e-05, "epoch": 1.8, "step": 48600}
+{"loss": 0.6451, "learning_rate": 2.992647877910215e-05, "epoch": 1.8, "step": 48610}
+{"loss": 0.64, "learning_rate": 2.9870780884482565e-05, "epoch": 1.8, "step": 48620}
+{"loss": 0.5696, "learning_rate": 2.981508298986298e-05, "epoch": 1.8, "step": 48630}
+{"loss": 0.5518, "learning_rate": 2.9759385095243397e-05, "epoch": 1.8, "step": 48640}
+{"loss": 0.5646, "learning_rate": 2.9703687200623815e-05, "epoch": 1.8, "step": 48650}
+{"loss": 0.4954, "learning_rate": 2.964798930600423e-05, "epoch": 1.8, "step": 48660}
+{"loss": 0.7372, "learning_rate": 2.9592291411384648e-05, "epoch": 1.8, "step": 48670}
+{"loss": 0.5751, "learning_rate": 2.9536593516765062e-05, "epoch": 1.8, "step": 48680}
+{"loss": 0.5941, "learning_rate": 2.9480895622145477e-05, "epoch": 1.8, "step": 48690}
+{"loss": 0.5785, "learning_rate": 2.9425197727525898e-05, "epoch": 1.8, "step": 48700}
+{"loss": 0.4956, "learning_rate": 2.9369499832906313e-05, "epoch": 1.81, "step": 48710}
+{"loss": 0.7268, "learning_rate": 2.931380193828673e-05, "epoch": 1.81, "step": 48720}
+{"loss": 0.5952, "learning_rate": 2.9258104043667145e-05, "epoch": 1.81, "step": 48730}
+{"loss": 0.6268, "learning_rate": 2.9202406149047566e-05, "epoch": 1.81, "step": 48740}
+{"loss": 0.4495, "learning_rate": 2.914670825442798e-05, "epoch": 1.81, "step": 48750}
+{"loss": 0.7017, "learning_rate": 2.9091010359808395e-05, "epoch": 1.81, "step": 48760}
+{"loss": 0.5829, "learning_rate": 2.9035312465188813e-05, "epoch": 1.81, "step": 48770}
+{"loss": 0.5505, "learning_rate": 2.897961457056923e-05, "epoch": 1.81, "step": 48780}
+{"loss": 0.5899, "learning_rate": 2.892391667594965e-05, "epoch": 1.81, "step": 48790}
+{"loss": 0.5719, "learning_rate": 2.8868218781330064e-05, "epoch": 1.81, "step": 48800}
+{"loss": 0.5271, "learning_rate": 2.8812520886710478e-05, "epoch": 1.81, "step": 48810}
+{"loss": 0.507, "learning_rate": 2.87568229920909e-05, "epoch": 1.81, "step": 48820}
+{"loss": 0.5156, "learning_rate": 2.8701125097471314e-05, "epoch": 1.81, "step": 48830}
+{"loss": 0.5059, "learning_rate": 2.864542720285173e-05, "epoch": 1.81, "step": 48840}
+{"loss": 0.5778, "learning_rate": 2.8589729308232146e-05, "epoch": 1.81, "step": 48850}
+{"loss": 0.6351, "learning_rate": 2.853403141361256e-05, "epoch": 1.81, "step": 48860}
+{"loss": 0.6452, "learning_rate": 2.8478333518992982e-05, "epoch": 1.81, "step": 48870}
+{"loss": 0.5668, "learning_rate": 2.8422635624373397e-05, "epoch": 1.81, "step": 48880}
+{"loss": 0.7163, "learning_rate": 2.836693772975381e-05, "epoch": 1.81, "step": 48890}
+{"loss": 0.5634, "learning_rate": 2.831123983513423e-05, "epoch": 1.81, "step": 48900}
+{"loss": 0.6061, "learning_rate": 2.8255541940514647e-05, "epoch": 1.81, "step": 48910}
+{"loss": 0.5385, "learning_rate": 2.8199844045895065e-05, "epoch": 1.81, "step": 48920}
+{"loss": 0.4921, "learning_rate": 2.814414615127548e-05, "epoch": 1.81, "step": 48930}
+{"loss": 0.502, "learning_rate": 2.8094018046117852e-05, "epoch": 1.81, "step": 48940}
+{"loss": 0.6084, "learning_rate": 2.8038320151498274e-05, "epoch": 1.81, "step": 48950}
+{"loss": 0.5424, "learning_rate": 2.7982622256878688e-05, "epoch": 1.81, "step": 48960}
+{"loss": 0.5544, "learning_rate": 2.7926924362259103e-05, "epoch": 1.81, "step": 48970}
+{"loss": 0.5997, "learning_rate": 2.787122646763952e-05, "epoch": 1.82, "step": 48980}
+{"loss": 0.642, "learning_rate": 2.781552857301994e-05, "epoch": 1.82, "step": 48990}
+{"loss": 0.5417, "learning_rate": 2.7759830678400356e-05, "epoch": 1.82, "step": 49000}
+{"loss": 0.5882, "learning_rate": 2.770413278378077e-05, "epoch": 1.82, "step": 49010}
+{"loss": 0.5367, "learning_rate": 2.7648434889161185e-05, "epoch": 1.82, "step": 49020}
+{"loss": 0.6134, "learning_rate": 2.7592736994541603e-05, "epoch": 1.82, "step": 49030}
+{"loss": 0.6281, "learning_rate": 2.753703909992202e-05, "epoch": 1.82, "step": 49040}
+{"loss": 0.5598, "learning_rate": 2.7481341205302436e-05, "epoch": 1.82, "step": 49050}
+{"loss": 0.4418, "learning_rate": 2.7425643310682854e-05, "epoch": 1.82, "step": 49060}
+{"loss": 0.5377, "learning_rate": 2.7369945416063268e-05, "epoch": 1.82, "step": 49070}
+{"loss": 0.5303, "learning_rate": 2.731424752144369e-05, "epoch": 1.82, "step": 49080}
+{"loss": 0.6383, "learning_rate": 2.7258549626824104e-05, "epoch": 1.82, "step": 49090}
+{"loss": 0.5223, "learning_rate": 2.720285173220452e-05, "epoch": 1.82, "step": 49100}
+{"loss": 0.5903, "learning_rate": 2.7147153837584936e-05, "epoch": 1.82, "step": 49110}
+{"loss": 0.5445, "learning_rate": 2.7091455942965354e-05, "epoch": 1.82, "step": 49120}
+{"loss": 0.5708, "learning_rate": 2.7035758048345772e-05, "epoch": 1.82, "step": 49130}
+{"loss": 0.5344, "learning_rate": 2.6980060153726187e-05, "epoch": 1.82, "step": 49140}
+{"loss": 0.5863, "learning_rate": 2.69243622591066e-05, "epoch": 1.82, "step": 49150}
+{"loss": 0.574, "learning_rate": 2.6868664364487023e-05, "epoch": 1.82, "step": 49160}
+{"loss": 0.4661, "learning_rate": 2.6812966469867437e-05, "epoch": 1.82, "step": 49170}
+{"loss": 0.5607, "learning_rate": 2.675726857524785e-05, "epoch": 1.82, "step": 49180}
+{"loss": 0.6135, "learning_rate": 2.670157068062827e-05, "epoch": 1.82, "step": 49190}
+{"loss": 0.5096, "learning_rate": 2.6645872786008684e-05, "epoch": 1.82, "step": 49200}
+{"loss": 0.6515, "learning_rate": 2.6590174891389105e-05, "epoch": 1.82, "step": 49210}
+{"loss": 0.6139, "learning_rate": 2.653447699676952e-05, "epoch": 1.82, "step": 49220}
+{"loss": 0.606, "learning_rate": 2.6478779102149934e-05, "epoch": 1.82, "step": 49230}
+{"loss": 0.5841, "learning_rate": 2.6423081207530352e-05, "epoch": 1.82, "step": 49240}
+{"loss": 0.5122, "learning_rate": 2.636738331291077e-05, "epoch": 1.83, "step": 49250}
+{"loss": 0.67, "learning_rate": 2.6311685418291188e-05, "epoch": 1.83, "step": 49260}
+{"loss": 0.6388, "learning_rate": 2.6255987523671603e-05, "epoch": 1.83, "step": 49270}
+{"loss": 0.5996, "learning_rate": 2.6200289629052017e-05, "epoch": 1.83, "step": 49280}
+{"loss": 0.5778, "learning_rate": 2.614459173443244e-05, "epoch": 1.83, "step": 49290}
+{"loss": 0.6016, "learning_rate": 2.6088893839812853e-05, "epoch": 1.83, "step": 49300}
+{"loss": 0.5354, "learning_rate": 2.6033195945193268e-05, "epoch": 1.83, "step": 49310}
+{"loss": 0.5299, "learning_rate": 2.5977498050573685e-05, "epoch": 1.83, "step": 49320}
+{"loss": 0.6332, "learning_rate": 2.5921800155954103e-05, "epoch": 1.83, "step": 49330}
+{"loss": 0.559, "learning_rate": 2.586610226133452e-05, "epoch": 1.83, "step": 49340}
+{"loss": 0.5202, "learning_rate": 2.5810404366714936e-05, "epoch": 1.83, "step": 49350}
+{"loss": 0.6037, "learning_rate": 2.575470647209535e-05, "epoch": 1.83, "step": 49360}
+{"loss": 0.5897, "learning_rate": 2.5699008577475768e-05, "epoch": 1.83, "step": 49370}
+{"loss": 0.4251, "learning_rate": 2.5643310682856186e-05, "epoch": 1.83, "step": 49380}
+{"loss": 0.5238, "learning_rate": 2.5587612788236604e-05, "epoch": 1.83, "step": 49390}
+{"loss": 0.65, "learning_rate": 2.553191489361702e-05, "epoch": 1.83, "step": 49400}
+{"loss": 0.4997, "learning_rate": 2.5476216998997433e-05, "epoch": 1.83, "step": 49410}
+{"loss": 0.5812, "learning_rate": 2.5420519104377854e-05, "epoch": 1.83, "step": 49420}
+{"loss": 0.529, "learning_rate": 2.536482120975827e-05, "epoch": 1.83, "step": 49430}
+{"loss": 0.447, "learning_rate": 2.5309123315138687e-05, "epoch": 1.83, "step": 49440}
+{"loss": 0.5349, "learning_rate": 2.52534254205191e-05, "epoch": 1.83, "step": 49450}
+{"loss": 0.6312, "learning_rate": 2.519772752589952e-05, "epoch": 1.83, "step": 49460}
+{"loss": 0.4897, "learning_rate": 2.5142029631279937e-05, "epoch": 1.83, "step": 49470}
+{"loss": 0.6419, "learning_rate": 2.5086331736660352e-05, "epoch": 1.83, "step": 49480}
+{"loss": 0.5988, "learning_rate": 2.5030633842040766e-05, "epoch": 1.83, "step": 49490}
+{"loss": 0.3872, "learning_rate": 2.4974935947421184e-05, "epoch": 1.83, "step": 49500}
+{"loss": 0.5554, "learning_rate": 2.4919238052801602e-05, "epoch": 1.83, "step": 49510}
+{"loss": 0.569, "learning_rate": 2.486354015818202e-05, "epoch": 1.84, "step": 49520}
+{"loss": 0.7036, "learning_rate": 2.4807842263562435e-05, "epoch": 1.84, "step": 49530}
+{"loss": 0.4347, "learning_rate": 2.475214436894285e-05, "epoch": 1.84, "step": 49540}
+{"loss": 0.5357, "learning_rate": 2.469644647432327e-05, "epoch": 1.84, "step": 49550}
+{"loss": 0.5829, "learning_rate": 2.4640748579703685e-05, "epoch": 1.84, "step": 49560}
+{"loss": 0.5407, "learning_rate": 2.4585050685084103e-05, "epoch": 1.84, "step": 49570}
+{"loss": 0.5562, "learning_rate": 2.4529352790464517e-05, "epoch": 1.84, "step": 49580}
+{"loss": 0.5497, "learning_rate": 2.4473654895844935e-05, "epoch": 1.84, "step": 49590}
+{"loss": 0.5843, "learning_rate": 2.4417957001225353e-05, "epoch": 1.84, "step": 49600}
+{"loss": 0.6302, "learning_rate": 2.4362259106605768e-05, "epoch": 1.84, "step": 49610}
+{"loss": 0.5279, "learning_rate": 2.4306561211986182e-05, "epoch": 1.84, "step": 49620}
+{"loss": 0.6376, "learning_rate": 2.4250863317366603e-05, "epoch": 1.84, "step": 49630}
+{"loss": 0.49, "learning_rate": 2.4195165422747018e-05, "epoch": 1.84, "step": 49640}
+{"loss": 0.5348, "learning_rate": 2.4139467528127436e-05, "epoch": 1.84, "step": 49650}
+{"loss": 0.6197, "learning_rate": 2.408376963350785e-05, "epoch": 1.84, "step": 49660}
+{"loss": 0.5806, "learning_rate": 2.4028071738888265e-05, "epoch": 1.84, "step": 49670}
+{"loss": 0.5545, "learning_rate": 2.3972373844268686e-05, "epoch": 1.84, "step": 49680}
+{"loss": 0.6144, "learning_rate": 2.39166759496491e-05, "epoch": 1.84, "step": 49690}
+{"loss": 0.4794, "learning_rate": 2.386097805502952e-05, "epoch": 1.84, "step": 49700}
+{"loss": 0.5592, "learning_rate": 2.3805280160409933e-05, "epoch": 1.84, "step": 49710}
+{"loss": 0.5108, "learning_rate": 2.374958226579035e-05, "epoch": 1.84, "step": 49720}
+{"loss": 0.4816, "learning_rate": 2.369388437117077e-05, "epoch": 1.84, "step": 49730}
+{"loss": 0.534, "learning_rate": 2.3638186476551184e-05, "epoch": 1.84, "step": 49740}
+{"loss": 0.7042, "learning_rate": 2.3582488581931598e-05, "epoch": 1.84, "step": 49750}
+{"loss": 0.6161, "learning_rate": 2.352679068731202e-05, "epoch": 1.84, "step": 49760}
+{"loss": 0.4955, "learning_rate": 2.3471092792692434e-05, "epoch": 1.84, "step": 49770}
+{"loss": 0.5406, "learning_rate": 2.3415394898072852e-05, "epoch": 1.85, "step": 49780}
+{"loss": 0.47, "learning_rate": 2.3359697003453266e-05, "epoch": 1.85, "step": 49790}
+{"loss": 0.5203, "learning_rate": 2.3303999108833688e-05, "epoch": 1.85, "step": 49800}
+{"loss": 0.4684, "learning_rate": 2.3248301214214102e-05, "epoch": 1.85, "step": 49810}
+{"loss": 0.64, "learning_rate": 2.3192603319594517e-05, "epoch": 1.85, "step": 49820}
+{"loss": 0.5931, "learning_rate": 2.3136905424974935e-05, "epoch": 1.85, "step": 49830}
+{"loss": 0.607, "learning_rate": 2.308120753035535e-05, "epoch": 1.85, "step": 49840}
+{"loss": 0.4432, "learning_rate": 2.3025509635735767e-05, "epoch": 1.85, "step": 49850}
+{"loss": 0.5324, "learning_rate": 2.2969811741116185e-05, "epoch": 1.85, "step": 49860}
+{"loss": 0.5889, "learning_rate": 2.29141138464966e-05, "epoch": 1.85, "step": 49870}
+{"loss": 0.4878, "learning_rate": 2.2858415951877014e-05, "epoch": 1.85, "step": 49880}
+{"loss": 0.5944, "learning_rate": 2.2802718057257435e-05, "epoch": 1.85, "step": 49890}
+{"loss": 0.6185, "learning_rate": 2.274702016263785e-05, "epoch": 1.85, "step": 49900}
+{"loss": 0.656, "learning_rate": 2.2691322268018268e-05, "epoch": 1.85, "step": 49910}
+{"loss": 0.5864, "learning_rate": 2.2635624373398682e-05, "epoch": 1.85, "step": 49920}
+{"loss": 0.4542, "learning_rate": 2.2579926478779104e-05, "epoch": 1.85, "step": 49930}
+{"loss": 0.6639, "learning_rate": 2.2524228584159518e-05, "epoch": 1.85, "step": 49940}
+{"loss": 0.7369, "learning_rate": 2.2468530689539933e-05, "epoch": 1.85, "step": 49950}
+{"loss": 0.5868, "learning_rate": 2.241283279492035e-05, "epoch": 1.85, "step": 49960}
+{"loss": 0.5113, "learning_rate": 2.2357134900300765e-05, "epoch": 1.85, "step": 49970}
+{"loss": 0.5578, "learning_rate": 2.2301437005681183e-05, "epoch": 1.85, "step": 49980}
+{"loss": 0.526, "learning_rate": 2.22457391110616e-05, "epoch": 1.85, "step": 49990}
+{"loss": 0.6395, "learning_rate": 2.2190041216442015e-05, "epoch": 1.85, "step": 50000}
+{"loss": 0.7001, "learning_rate": 2.213434332182243e-05, "epoch": 1.85, "step": 50010}
+{"loss": 0.4927, "learning_rate": 2.207864542720285e-05, "epoch": 1.85, "step": 50020}
+{"loss": 0.58, "learning_rate": 2.2022947532583266e-05, "epoch": 1.85, "step": 50030}
+{"loss": 0.5076, "learning_rate": 2.1967249637963684e-05, "epoch": 1.85, "step": 50040}
+{"loss": 0.5777, "learning_rate": 2.1911551743344098e-05, "epoch": 1.86, "step": 50050}
+{"loss": 0.6302, "learning_rate": 2.185585384872452e-05, "epoch": 1.86, "step": 50060}
+{"loss": 0.5982, "learning_rate": 2.1800155954104934e-05, "epoch": 1.86, "step": 50070}
+{"loss": 0.4589, "learning_rate": 2.174445805948535e-05, "epoch": 1.86, "step": 50080}
+{"loss": 0.5075, "learning_rate": 2.1688760164865766e-05, "epoch": 1.86, "step": 50090}
+{"loss": 0.5355, "learning_rate": 2.1633062270246184e-05, "epoch": 1.86, "step": 50100}
+{"loss": 0.5174, "learning_rate": 2.15773643756266e-05, "epoch": 1.86, "step": 50110}
+{"loss": 0.6385, "learning_rate": 2.1521666481007017e-05, "epoch": 1.86, "step": 50120}
+{"loss": 0.648, "learning_rate": 2.146596858638743e-05, "epoch": 1.86, "step": 50130}
+{"loss": 0.4629, "learning_rate": 2.1410270691767846e-05, "epoch": 1.86, "step": 50140}
+{"loss": 0.5356, "learning_rate": 2.1354572797148267e-05, "epoch": 1.86, "step": 50150}
+{"loss": 0.589, "learning_rate": 2.129887490252868e-05, "epoch": 1.86, "step": 50160}
+{"loss": 0.485, "learning_rate": 2.12431770079091e-05, "epoch": 1.86, "step": 50170}
+{"loss": 0.6254, "learning_rate": 2.1187479113289514e-05, "epoch": 1.86, "step": 50180}
+{"loss": 0.4952, "learning_rate": 2.1131781218669935e-05, "epoch": 1.86, "step": 50190}
+{"loss": 0.5772, "learning_rate": 2.107608332405035e-05, "epoch": 1.86, "step": 50200}
+{"loss": 0.5264, "learning_rate": 2.1020385429430764e-05, "epoch": 1.86, "step": 50210}
+{"loss": 0.448, "learning_rate": 2.0964687534811182e-05, "epoch": 1.86, "step": 50220}
+{"loss": 0.523, "learning_rate": 2.09089896401916e-05, "epoch": 1.86, "step": 50230}
+{"loss": 0.5539, "learning_rate": 2.0853291745572015e-05, "epoch": 1.86, "step": 50240}
+{"loss": 0.5861, "learning_rate": 2.0797593850952433e-05, "epoch": 1.86, "step": 50250}
+{"loss": 0.5885, "learning_rate": 2.0741895956332847e-05, "epoch": 1.86, "step": 50260}
+{"loss": 0.6078, "learning_rate": 2.068619806171327e-05, "epoch": 1.86, "step": 50270}
+{"loss": 0.6467, "learning_rate": 2.0630500167093683e-05, "epoch": 1.86, "step": 50280}
+{"loss": 0.5114, "learning_rate": 2.0574802272474098e-05, "epoch": 1.86, "step": 50290}
+{"loss": 0.5043, "learning_rate": 2.0519104377854516e-05, "epoch": 1.86, "step": 50300}
+{"loss": 0.7764, "learning_rate": 2.046340648323493e-05, "epoch": 1.86, "step": 50310}
+{"loss": 0.5989, "learning_rate": 2.040770858861535e-05, "epoch": 1.87, "step": 50320}
+{"loss": 0.5397, "learning_rate": 2.0352010693995766e-05, "epoch": 1.87, "step": 50330}
+{"loss": 0.6624, "learning_rate": 2.029631279937618e-05, "epoch": 1.87, "step": 50340}
+{"loss": 0.5074, "learning_rate": 2.0240614904756598e-05, "epoch": 1.87, "step": 50350}
+{"loss": 0.6125, "learning_rate": 2.0184917010137016e-05, "epoch": 1.87, "step": 50360}
+{"loss": 0.6758, "learning_rate": 2.012921911551743e-05, "epoch": 1.87, "step": 50370}
+{"loss": 0.7117, "learning_rate": 2.007352122089785e-05, "epoch": 1.87, "step": 50380}
+{"loss": 0.7379, "learning_rate": 2.0017823326278263e-05, "epoch": 1.87, "step": 50390}
+{"loss": 0.4451, "learning_rate": 1.9962125431658684e-05, "epoch": 1.87, "step": 50400}
+{"loss": 0.4612, "learning_rate": 1.99064275370391e-05, "epoch": 1.87, "step": 50410}
+{"loss": 0.5497, "learning_rate": 1.9850729642419514e-05, "epoch": 1.87, "step": 50420}
+{"loss": 0.6251, "learning_rate": 1.979503174779993e-05, "epoch": 1.87, "step": 50430}
+{"loss": 0.5437, "learning_rate": 1.9739333853180346e-05, "epoch": 1.87, "step": 50440}
+{"loss": 0.5941, "learning_rate": 1.9683635958560767e-05, "epoch": 1.87, "step": 50450}
+{"loss": 0.5474, "learning_rate": 1.9627938063941182e-05, "epoch": 1.87, "step": 50460}
+{"loss": 0.5998, "learning_rate": 1.9572240169321596e-05, "epoch": 1.87, "step": 50470}
+{"loss": 0.5055, "learning_rate": 1.9516542274702014e-05, "epoch": 1.87, "step": 50480}
+{"loss": 0.5535, "learning_rate": 1.9460844380082432e-05, "epoch": 1.87, "step": 50490}
+{"loss": 0.5476, "learning_rate": 1.9405146485462847e-05, "epoch": 1.87, "step": 50500}
+{"loss": 0.5119, "learning_rate": 1.9349448590843265e-05, "epoch": 1.87, "step": 50510}
+{"loss": 0.5713, "learning_rate": 1.929375069622368e-05, "epoch": 1.87, "step": 50520}
+{"loss": 0.4773, "learning_rate": 1.92380528016041e-05, "epoch": 1.87, "step": 50530}
+{"loss": 0.5768, "learning_rate": 1.9182354906984515e-05, "epoch": 1.87, "step": 50540}
+{"loss": 0.5221, "learning_rate": 1.912665701236493e-05, "epoch": 1.87, "step": 50550}
+{"loss": 0.7206, "learning_rate": 1.9070959117745347e-05, "epoch": 1.87, "step": 50560}
+{"loss": 0.4983, "learning_rate": 1.9015261223125765e-05, "epoch": 1.87, "step": 50570}
+{"loss": 0.5234, "learning_rate": 1.8959563328506183e-05, "epoch": 1.87, "step": 50580}
+{"loss": 0.521, "learning_rate": 1.8903865433886598e-05, "epoch": 1.88, "step": 50590}
+{"loss": 0.5795, "learning_rate": 1.8848167539267012e-05, "epoch": 1.88, "step": 50600}
+{"loss": 0.6918, "learning_rate": 1.879246964464743e-05, "epoch": 1.88, "step": 50610}
+{"loss": 0.6866, "learning_rate": 1.8736771750027848e-05, "epoch": 1.88, "step": 50620}
+{"loss": 0.6993, "learning_rate": 1.8681073855408263e-05, "epoch": 1.88, "step": 50630}
+{"loss": 0.6497, "learning_rate": 1.862537596078868e-05, "epoch": 1.88, "step": 50640}
+{"loss": 0.4933, "learning_rate": 1.85696780661691e-05, "epoch": 1.88, "step": 50650}
+{"loss": 0.6868, "learning_rate": 1.8513980171549513e-05, "epoch": 1.88, "step": 50660}
+{"loss": 0.5818, "learning_rate": 1.845828227692993e-05, "epoch": 1.88, "step": 50670}
+{"loss": 0.5683, "learning_rate": 1.8402584382310345e-05, "epoch": 1.88, "step": 50680}
+{"loss": 0.4418, "learning_rate": 1.8346886487690763e-05, "epoch": 1.88, "step": 50690}
+{"loss": 0.5005, "learning_rate": 1.829118859307118e-05, "epoch": 1.88, "step": 50700}
+{"loss": 0.5944, "learning_rate": 1.82354906984516e-05, "epoch": 1.88, "step": 50710}
+{"loss": 0.7682, "learning_rate": 1.8179792803832014e-05, "epoch": 1.88, "step": 50720}
+{"loss": 0.4511, "learning_rate": 1.812409490921243e-05, "epoch": 1.88, "step": 50730}
+{"loss": 0.5313, "learning_rate": 1.8068397014592846e-05, "epoch": 1.88, "step": 50740}
+{"loss": 0.5761, "learning_rate": 1.8012699119973264e-05, "epoch": 1.88, "step": 50750}
+{"loss": 0.6557, "learning_rate": 1.7957001225353682e-05, "epoch": 1.88, "step": 50760}
+{"loss": 0.5167, "learning_rate": 1.7901303330734096e-05, "epoch": 1.88, "step": 50770}
+{"loss": 0.5412, "learning_rate": 1.7845605436114514e-05, "epoch": 1.88, "step": 50780}
+{"loss": 0.5521, "learning_rate": 1.778990754149493e-05, "epoch": 1.88, "step": 50790}
+{"loss": 0.537, "learning_rate": 1.7734209646875347e-05, "epoch": 1.88, "step": 50800}
+{"loss": 0.5333, "learning_rate": 1.767851175225576e-05, "epoch": 1.88, "step": 50810}
+{"loss": 0.5173, "learning_rate": 1.762281385763618e-05, "epoch": 1.88, "step": 50820}
+{"loss": 0.4469, "learning_rate": 1.7567115963016597e-05, "epoch": 1.88, "step": 50830}
+{"loss": 0.5463, "learning_rate": 1.7511418068397015e-05, "epoch": 1.88, "step": 50840}
+{"loss": 0.5388, "learning_rate": 1.745572017377743e-05, "epoch": 1.88, "step": 50850}
+{"loss": 0.6881, "learning_rate": 1.7400022279157847e-05, "epoch": 1.89, "step": 50860}
+{"loss": 0.5108, "learning_rate": 1.7344324384538262e-05, "epoch": 1.89, "step": 50870}
+{"loss": 0.4259, "learning_rate": 1.728862648991868e-05, "epoch": 1.89, "step": 50880}
+{"loss": 0.5143, "learning_rate": 1.7232928595299098e-05, "epoch": 1.89, "step": 50890}
+{"loss": 0.4401, "learning_rate": 1.7177230700679512e-05, "epoch": 1.89, "step": 50900}
+{"loss": 0.5805, "learning_rate": 1.712153280605993e-05, "epoch": 1.89, "step": 50910}
+{"loss": 0.5834, "learning_rate": 1.7065834911440345e-05, "epoch": 1.89, "step": 50920}
+{"loss": 0.6101, "learning_rate": 1.7010137016820763e-05, "epoch": 1.89, "step": 50930}
+{"loss": 0.6045, "learning_rate": 1.6954439122201177e-05, "epoch": 1.89, "step": 50940}
+{"loss": 0.5697, "learning_rate": 1.6898741227581595e-05, "epoch": 1.89, "step": 50950}
+{"loss": 0.5748, "learning_rate": 1.6843043332962013e-05, "epoch": 1.89, "step": 50960}
+{"loss": 0.5727, "learning_rate": 1.678734543834243e-05, "epoch": 1.89, "step": 50970}
+{"loss": 0.4418, "learning_rate": 1.6731647543722845e-05, "epoch": 1.89, "step": 50980}
+{"loss": 0.6387, "learning_rate": 1.6675949649103263e-05, "epoch": 1.89, "step": 50990}
+{"loss": 0.5258, "learning_rate": 1.6620251754483678e-05, "epoch": 1.89, "step": 51000}
+{"loss": 0.4365, "learning_rate": 1.6564553859864096e-05, "epoch": 1.89, "step": 51010}
+{"loss": 0.5777, "learning_rate": 1.6508855965244514e-05, "epoch": 1.89, "step": 51020}
+{"loss": 0.5431, "learning_rate": 1.6453158070624928e-05, "epoch": 1.89, "step": 51030}
+{"loss": 0.6415, "learning_rate": 1.6397460176005346e-05, "epoch": 1.89, "step": 51040}
+{"loss": 0.5364, "learning_rate": 1.6341762281385764e-05, "epoch": 1.89, "step": 51050}
+{"loss": 0.5547, "learning_rate": 1.628606438676618e-05, "epoch": 1.89, "step": 51060}
+{"loss": 0.531, "learning_rate": 1.6230366492146593e-05, "epoch": 1.89, "step": 51070}
+{"loss": 0.5839, "learning_rate": 1.617466859752701e-05, "epoch": 1.89, "step": 51080}
+{"loss": 0.6383, "learning_rate": 1.611897070290743e-05, "epoch": 1.89, "step": 51090}
+{"loss": 0.4692, "learning_rate": 1.6063272808287847e-05, "epoch": 1.89, "step": 51100}
+{"loss": 0.4786, "learning_rate": 1.600757491366826e-05, "epoch": 1.89, "step": 51110}
+{"loss": 0.741, "learning_rate": 1.595187701904868e-05, "epoch": 1.89, "step": 51120}
+{"loss": 0.5523, "learning_rate": 1.5896179124429094e-05, "epoch": 1.9, "step": 51130}
+{"loss": 0.6348, "learning_rate": 1.5840481229809512e-05, "epoch": 1.9, "step": 51140}
+{"loss": 0.5444, "learning_rate": 1.578478333518993e-05, "epoch": 1.9, "step": 51150}
+{"loss": 0.551, "learning_rate": 1.5729085440570344e-05, "epoch": 1.9, "step": 51160}
+{"loss": 0.5785, "learning_rate": 1.5673387545950762e-05, "epoch": 1.9, "step": 51170}
+{"loss": 0.6517, "learning_rate": 1.561768965133118e-05, "epoch": 1.9, "step": 51180}
+{"loss": 0.4782, "learning_rate": 1.5561991756711594e-05, "epoch": 1.9, "step": 51190}
+{"loss": 0.6313, "learning_rate": 1.5506293862092012e-05, "epoch": 1.9, "step": 51200}
+{"loss": 0.6293, "learning_rate": 1.5450595967472427e-05, "epoch": 1.9, "step": 51210}
+{"loss": 0.5385, "learning_rate": 1.5394898072852845e-05, "epoch": 1.9, "step": 51220}
+{"loss": 0.5399, "learning_rate": 1.5339200178233263e-05, "epoch": 1.9, "step": 51230}
+{"loss": 0.6111, "learning_rate": 1.5283502283613677e-05, "epoch": 1.9, "step": 51240}
+{"loss": 0.7043, "learning_rate": 1.5227804388994095e-05, "epoch": 1.9, "step": 51250}
+{"loss": 0.4684, "learning_rate": 1.5172106494374511e-05, "epoch": 1.9, "step": 51260}
+{"loss": 0.5684, "learning_rate": 1.5116408599754928e-05, "epoch": 1.9, "step": 51270}
+{"loss": 0.3748, "learning_rate": 1.5060710705135344e-05, "epoch": 1.9, "step": 51280}
+{"loss": 0.6251, "learning_rate": 1.5005012810515762e-05, "epoch": 1.9, "step": 51290}
+{"loss": 0.5439, "learning_rate": 1.4949314915896178e-05, "epoch": 1.9, "step": 51300}
+{"loss": 0.5666, "learning_rate": 1.4893617021276596e-05, "epoch": 1.9, "step": 51310}
+{"loss": 0.6856, "learning_rate": 1.483791912665701e-05, "epoch": 1.9, "step": 51320}
+{"loss": 0.5606, "learning_rate": 1.4782221232037428e-05, "epoch": 1.9, "step": 51330}
+{"loss": 0.5947, "learning_rate": 1.4726523337417845e-05, "epoch": 1.9, "step": 51340}
+{"loss": 0.5522, "learning_rate": 1.4670825442798262e-05, "epoch": 1.9, "step": 51350}
+{"loss": 0.5981, "learning_rate": 1.4615127548178677e-05, "epoch": 1.9, "step": 51360}
+{"loss": 0.5534, "learning_rate": 1.4559429653559093e-05, "epoch": 1.9, "step": 51370}
+{"loss": 0.4942, "learning_rate": 1.4503731758939511e-05, "epoch": 1.9, "step": 51380}
+{"loss": 0.4048, "learning_rate": 1.4448033864319927e-05, "epoch": 1.9, "step": 51390}
+{"loss": 0.4822, "learning_rate": 1.4392335969700344e-05, "epoch": 1.91, "step": 51400}
+{"loss": 0.5823, "learning_rate": 1.433663807508076e-05, "epoch": 1.91, "step": 51410}
+{"loss": 0.6042, "learning_rate": 1.4280940180461178e-05, "epoch": 1.91, "step": 51420}
+{"loss": 0.5983, "learning_rate": 1.4225242285841594e-05, "epoch": 1.91, "step": 51430}
+{"loss": 0.4227, "learning_rate": 1.4169544391222012e-05, "epoch": 1.91, "step": 51440}
+{"loss": 0.4886, "learning_rate": 1.4113846496602426e-05, "epoch": 1.91, "step": 51450}
+{"loss": 0.5644, "learning_rate": 1.4058148601982844e-05, "epoch": 1.91, "step": 51460}
+{"loss": 0.5254, "learning_rate": 1.400245070736326e-05, "epoch": 1.91, "step": 51470}
+{"loss": 0.4171, "learning_rate": 1.3946752812743678e-05, "epoch": 1.91, "step": 51480}
+{"loss": 0.586, "learning_rate": 1.3891054918124093e-05, "epoch": 1.91, "step": 51490}
+{"loss": 0.4959, "learning_rate": 1.383535702350451e-05, "epoch": 1.91, "step": 51500}
+{"loss": 0.6238, "learning_rate": 1.3779659128884927e-05, "epoch": 1.91, "step": 51510}
+{"loss": 0.53, "learning_rate": 1.3723961234265345e-05, "epoch": 1.91, "step": 51520}
+{"loss": 0.428, "learning_rate": 1.366826333964576e-05, "epoch": 1.91, "step": 51530}
+{"loss": 0.576, "learning_rate": 1.3612565445026176e-05, "epoch": 1.91, "step": 51540}
+{"loss": 0.5181, "learning_rate": 1.3556867550406594e-05, "epoch": 1.91, "step": 51550}
+{"loss": 0.5524, "learning_rate": 1.350116965578701e-05, "epoch": 1.91, "step": 51560}
+{"loss": 0.4988, "learning_rate": 1.3445471761167428e-05, "epoch": 1.91, "step": 51570}
+{"loss": 0.5608, "learning_rate": 1.3389773866547842e-05, "epoch": 1.91, "step": 51580}
+{"loss": 0.5204, "learning_rate": 1.333407597192826e-05, "epoch": 1.91, "step": 51590}
+{"loss": 0.5184, "learning_rate": 1.3278378077308676e-05, "epoch": 1.91, "step": 51600}
+{"loss": 0.5935, "learning_rate": 1.3222680182689094e-05, "epoch": 1.91, "step": 51610}
+{"loss": 0.4812, "learning_rate": 1.3166982288069509e-05, "epoch": 1.91, "step": 51620}
+{"loss": 0.6615, "learning_rate": 1.3111284393449927e-05, "epoch": 1.91, "step": 51630}
+{"loss": 0.5362, "learning_rate": 1.3055586498830343e-05, "epoch": 1.91, "step": 51640}
+{"loss": 0.6029, "learning_rate": 1.2999888604210761e-05, "epoch": 1.91, "step": 51650}
+{"loss": 0.698, "learning_rate": 1.2944190709591175e-05, "epoch": 1.91, "step": 51660}
+{"loss": 0.5691, "learning_rate": 1.2888492814971593e-05, "epoch": 1.92, "step": 51670}
+{"loss": 0.5093, "learning_rate": 1.283279492035201e-05, "epoch": 1.92, "step": 51680}
+{"loss": 0.6178, "learning_rate": 1.2777097025732426e-05, "epoch": 1.92, "step": 51690}
+{"loss": 0.4218, "learning_rate": 1.2721399131112844e-05, "epoch": 1.92, "step": 51700}
+{"loss": 0.6245, "learning_rate": 1.2665701236493258e-05, "epoch": 1.92, "step": 51710}
+{"loss": 0.5696, "learning_rate": 1.2610003341873676e-05, "epoch": 1.92, "step": 51720}
+{"loss": 0.552, "learning_rate": 1.2554305447254092e-05, "epoch": 1.92, "step": 51730}
+{"loss": 0.5483, "learning_rate": 1.249860755263451e-05, "epoch": 1.92, "step": 51740}
+{"loss": 0.5139, "learning_rate": 1.2442909658014925e-05, "epoch": 1.92, "step": 51750}
+{"loss": 0.5659, "learning_rate": 1.2387211763395343e-05, "epoch": 1.92, "step": 51760}
+{"loss": 0.557, "learning_rate": 1.2331513868775759e-05, "epoch": 1.92, "step": 51770}
+{"loss": 0.622, "learning_rate": 1.2275815974156177e-05, "epoch": 1.92, "step": 51780}
+{"loss": 0.5288, "learning_rate": 1.2220118079536593e-05, "epoch": 1.92, "step": 51790}
+{"loss": 0.5918, "learning_rate": 1.216442018491701e-05, "epoch": 1.92, "step": 51800}
+{"loss": 0.4625, "learning_rate": 1.2108722290297425e-05, "epoch": 1.92, "step": 51810}
+{"loss": 0.7206, "learning_rate": 1.2053024395677843e-05, "epoch": 1.92, "step": 51820}
+{"loss": 0.4559, "learning_rate": 1.199732650105826e-05, "epoch": 1.92, "step": 51830}
+{"loss": 0.5578, "learning_rate": 1.1941628606438674e-05, "epoch": 1.92, "step": 51840}
+{"loss": 0.6598, "learning_rate": 1.1885930711819092e-05, "epoch": 1.92, "step": 51850}
+{"loss": 0.5262, "learning_rate": 1.1830232817199508e-05, "epoch": 1.92, "step": 51860}
+{"loss": 0.5228, "learning_rate": 1.1774534922579926e-05, "epoch": 1.92, "step": 51870}
+{"loss": 0.5485, "learning_rate": 1.171883702796034e-05, "epoch": 1.92, "step": 51880}
+{"loss": 0.5205, "learning_rate": 1.1663139133340759e-05, "epoch": 1.92, "step": 51890}
+{"loss": 0.5783, "learning_rate": 1.1607441238721175e-05, "epoch": 1.92, "step": 51900}
+{"loss": 0.5976, "learning_rate": 1.1551743344101593e-05, "epoch": 1.92, "step": 51910}
+{"loss": 0.5747, "learning_rate": 1.1496045449482009e-05, "epoch": 1.92, "step": 51920}
+{"loss": 0.6186, "learning_rate": 1.1440347554862425e-05, "epoch": 1.92, "step": 51930}
+{"loss": 0.4475, "learning_rate": 1.1384649660242841e-05, "epoch": 1.93, "step": 51940}
+{"loss": 0.8409, "learning_rate": 1.132895176562326e-05, "epoch": 1.93, "step": 51950}
+{"loss": 0.5911, "learning_rate": 1.1273253871003675e-05, "epoch": 1.93, "step": 51960}
+{"loss": 0.6518, "learning_rate": 1.1217555976384093e-05, "epoch": 1.93, "step": 51970}
+{"loss": 0.5714, "learning_rate": 1.1161858081764508e-05, "epoch": 1.93, "step": 51980}
+{"loss": 0.5216, "learning_rate": 1.1106160187144924e-05, "epoch": 1.93, "step": 51990}
+{"loss": 0.5451, "learning_rate": 1.1050462292525342e-05, "epoch": 1.93, "step": 52000}
+{"loss": 0.5307, "learning_rate": 1.0994764397905757e-05, "epoch": 1.93, "step": 52010}
+{"loss": 0.5078, "learning_rate": 1.0939066503286174e-05, "epoch": 1.93, "step": 52020}
+{"loss": 0.7074, "learning_rate": 1.088336860866659e-05, "epoch": 1.93, "step": 52030}
+{"loss": 0.4657, "learning_rate": 1.0827670714047009e-05, "epoch": 1.93, "step": 52040}
+{"loss": 0.6012, "learning_rate": 1.0771972819427425e-05, "epoch": 1.93, "step": 52050}
+{"loss": 0.5066, "learning_rate": 1.0716274924807841e-05, "epoch": 1.93, "step": 52060}
+{"loss": 0.4725, "learning_rate": 1.0660577030188257e-05, "epoch": 1.93, "step": 52070}
+{"loss": 0.5967, "learning_rate": 1.0604879135568675e-05, "epoch": 1.93, "step": 52080}
+{"loss": 0.4897, "learning_rate": 1.0549181240949091e-05, "epoch": 1.93, "step": 52090}
+{"loss": 0.5391, "learning_rate": 1.049348334632951e-05, "epoch": 1.93, "step": 52100}
+{"loss": 0.4885, "learning_rate": 1.0437785451709924e-05, "epoch": 1.93, "step": 52110}
+{"loss": 0.577, "learning_rate": 1.0382087557090342e-05, "epoch": 1.93, "step": 52120}
+{"loss": 0.5491, "learning_rate": 1.0326389662470758e-05, "epoch": 1.93, "step": 52130}
+{"loss": 0.5647, "learning_rate": 1.0270691767851176e-05, "epoch": 1.93, "step": 52140}
+{"loss": 0.4779, "learning_rate": 1.021499387323159e-05, "epoch": 1.93, "step": 52150}
+{"loss": 0.5625, "learning_rate": 1.0159295978612007e-05, "epoch": 1.93, "step": 52160}
+{"loss": 0.555, "learning_rate": 1.0103598083992425e-05, "epoch": 1.93, "step": 52170}
+{"loss": 0.5129, "learning_rate": 1.004790018937284e-05, "epoch": 1.93, "step": 52180}
+{"loss": 0.5839, "learning_rate": 9.992202294753257e-06, "epoch": 1.93, "step": 52190}
+{"loss": 0.5261, "learning_rate": 9.936504400133673e-06, "epoch": 1.93, "step": 52200}
+{"loss": 0.6553, "learning_rate": 9.880806505514091e-06, "epoch": 1.94, "step": 52210}
+{"loss": 0.4706, "learning_rate": 9.825108610894507e-06, "epoch": 1.94, "step": 52220}
+{"loss": 0.6074, "learning_rate": 9.769410716274925e-06, "epoch": 1.94, "step": 52230}
+{"loss": 0.4789, "learning_rate": 9.71371282165534e-06, "epoch": 1.94, "step": 52240}
+{"loss": 0.5131, "learning_rate": 9.658014927035758e-06, "epoch": 1.94, "step": 52250}
+{"loss": 0.5238, "learning_rate": 9.602317032416174e-06, "epoch": 1.94, "step": 52260}
+{"loss": 0.606, "learning_rate": 9.546619137796592e-06, "epoch": 1.94, "step": 52270}
+{"loss": 0.6541, "learning_rate": 9.490921243177006e-06, "epoch": 1.94, "step": 52280}
+{"loss": 0.6714, "learning_rate": 9.435223348557424e-06, "epoch": 1.94, "step": 52290}
+{"loss": 0.6529, "learning_rate": 9.37952545393784e-06, "epoch": 1.94, "step": 52300}
+{"loss": 0.5762, "learning_rate": 9.323827559318257e-06, "epoch": 1.94, "step": 52310}
+{"loss": 0.5856, "learning_rate": 9.268129664698673e-06, "epoch": 1.94, "step": 52320}
+{"loss": 0.5428, "learning_rate": 9.21243177007909e-06, "epoch": 1.94, "step": 52330}
+{"loss": 0.5138, "learning_rate": 9.156733875459507e-06, "epoch": 1.94, "step": 52340}
+{"loss": 0.4142, "learning_rate": 9.101035980839923e-06, "epoch": 1.94, "step": 52350}
+{"loss": 0.5272, "learning_rate": 9.045338086220341e-06, "epoch": 1.94, "step": 52360}
+{"loss": 0.5971, "learning_rate": 8.989640191600757e-06, "epoch": 1.94, "step": 52370}
+{"loss": 0.6514, "learning_rate": 8.933942296981174e-06, "epoch": 1.94, "step": 52380}
+{"loss": 0.5154, "learning_rate": 8.87824440236159e-06, "epoch": 1.94, "step": 52390}
+{"loss": 0.4522, "learning_rate": 8.822546507742006e-06, "epoch": 1.94, "step": 52400}
+{"loss": 0.7203, "learning_rate": 8.766848613122422e-06, "epoch": 1.94, "step": 52410}
+{"loss": 0.4902, "learning_rate": 8.71115071850284e-06, "epoch": 1.94, "step": 52420}
+{"loss": 0.4589, "learning_rate": 8.655452823883256e-06, "epoch": 1.94, "step": 52430}
+{"loss": 0.5249, "learning_rate": 8.599754929263673e-06, "epoch": 1.94, "step": 52440}
+{"loss": 0.571, "learning_rate": 8.54405703464409e-06, "epoch": 1.94, "step": 52450}
+{"loss": 0.5981, "learning_rate": 8.488359140024507e-06, "epoch": 1.94, "step": 52460}
+{"loss": 0.5864, "learning_rate": 8.432661245404923e-06, "epoch": 1.94, "step": 52470}
+{"loss": 0.625, "learning_rate": 8.376963350785339e-06, "epoch": 1.95, "step": 52480}
+{"loss": 0.7408, "learning_rate": 8.321265456165757e-06, "epoch": 1.95, "step": 52490}
+{"loss": 0.5908, "learning_rate": 8.265567561546173e-06, "epoch": 1.95, "step": 52500}
+{"loss": 0.558, "learning_rate": 8.20986966692659e-06, "epoch": 1.95, "step": 52510}
+{"loss": 0.6495, "learning_rate": 8.154171772307006e-06, "epoch": 1.95, "step": 52520}
+{"loss": 0.6201, "learning_rate": 8.098473877687424e-06, "epoch": 1.95, "step": 52530}
+{"loss": 0.4979, "learning_rate": 8.042775983067838e-06, "epoch": 1.95, "step": 52540}
+{"loss": 0.605, "learning_rate": 7.987078088448256e-06, "epoch": 1.95, "step": 52550}
+{"loss": 0.4532, "learning_rate": 7.931380193828672e-06, "epoch": 1.95, "step": 52560}
+{"loss": 0.5578, "learning_rate": 7.875682299209088e-06, "epoch": 1.95, "step": 52570}
+{"loss": 0.5913, "learning_rate": 7.819984404589506e-06, "epoch": 1.95, "step": 52580}
+{"loss": 0.6065, "learning_rate": 7.764286509969923e-06, "epoch": 1.95, "step": 52590}
+{"loss": 0.575, "learning_rate": 7.708588615350339e-06, "epoch": 1.95, "step": 52600}
+{"loss": 0.5402, "learning_rate": 7.652890720730755e-06, "epoch": 1.95, "step": 52610}
+{"loss": 0.5447, "learning_rate": 7.597192826111172e-06, "epoch": 1.95, "step": 52620}
+{"loss": 0.7365, "learning_rate": 7.541494931491589e-06, "epoch": 1.95, "step": 52630}
+{"loss": 0.5948, "learning_rate": 7.485797036872005e-06, "epoch": 1.95, "step": 52640}
+{"loss": 0.613, "learning_rate": 7.4300991422524225e-06, "epoch": 1.95, "step": 52650}
+{"loss": 0.5444, "learning_rate": 7.3744012476328395e-06, "epoch": 1.95, "step": 52660}
+{"loss": 0.6483, "learning_rate": 7.318703353013256e-06, "epoch": 1.95, "step": 52670}
+{"loss": 0.5607, "learning_rate": 7.263005458393673e-06, "epoch": 1.95, "step": 52680}
+{"loss": 0.438, "learning_rate": 7.207307563774088e-06, "epoch": 1.95, "step": 52690}
+{"loss": 0.555, "learning_rate": 7.151609669154505e-06, "epoch": 1.95, "step": 52700}
+{"loss": 0.7265, "learning_rate": 7.0959117745349215e-06, "epoch": 1.95, "step": 52710}
+{"loss": 0.559, "learning_rate": 7.0402138799153385e-06, "epoch": 1.95, "step": 52720}
+{"loss": 0.6297, "learning_rate": 6.984515985295755e-06, "epoch": 1.95, "step": 52730}
+{"loss": 0.6368, "learning_rate": 6.928818090676172e-06, "epoch": 1.95, "step": 52740}
+{"loss": 0.6207, "learning_rate": 6.873120196056588e-06, "epoch": 1.96, "step": 52750}
+{"loss": 0.6341, "learning_rate": 6.817422301437005e-06, "epoch": 1.96, "step": 52760}
+{"loss": 0.491, "learning_rate": 6.761724406817421e-06, "epoch": 1.96, "step": 52770}
+{"loss": 0.4479, "learning_rate": 6.706026512197838e-06, "epoch": 1.96, "step": 52780}
+{"loss": 0.5128, "learning_rate": 6.6503286175782554e-06, "epoch": 1.96, "step": 52790}
+{"loss": 0.4408, "learning_rate": 6.594630722958672e-06, "epoch": 1.96, "step": 52800}
+{"loss": 0.5033, "learning_rate": 6.538932828339089e-06, "epoch": 1.96, "step": 52810}
+{"loss": 0.6035, "learning_rate": 6.483234933719505e-06, "epoch": 1.96, "step": 52820}
+{"loss": 0.501, "learning_rate": 6.427537039099922e-06, "epoch": 1.96, "step": 52830}
+{"loss": 0.5874, "learning_rate": 6.371839144480338e-06, "epoch": 1.96, "step": 52840}
+{"loss": 0.4779, "learning_rate": 6.3161412498607545e-06, "epoch": 1.96, "step": 52850}
+{"loss": 0.6621, "learning_rate": 6.260443355241171e-06, "epoch": 1.96, "step": 52860}
+{"loss": 0.5001, "learning_rate": 6.204745460621588e-06, "epoch": 1.96, "step": 52870}
+{"loss": 0.5606, "learning_rate": 6.149047566002004e-06, "epoch": 1.96, "step": 52880}
+{"loss": 0.566, "learning_rate": 6.093349671382421e-06, "epoch": 1.96, "step": 52890}
+{"loss": 0.553, "learning_rate": 6.037651776762838e-06, "epoch": 1.96, "step": 52900}
+{"loss": 0.5356, "learning_rate": 5.981953882143254e-06, "epoch": 1.96, "step": 52910}
+{"loss": 0.5597, "learning_rate": 5.926255987523671e-06, "epoch": 1.96, "step": 52920}
+{"loss": 0.5982, "learning_rate": 5.870558092904088e-06, "epoch": 1.96, "step": 52930}
+{"loss": 0.5086, "learning_rate": 5.814860198284505e-06, "epoch": 1.96, "step": 52940}
+{"loss": 0.4721, "learning_rate": 5.759162303664921e-06, "epoch": 1.96, "step": 52950}
+{"loss": 0.7, "learning_rate": 5.703464409045338e-06, "epoch": 1.96, "step": 52960}
+{"loss": 0.4575, "learning_rate": 5.647766514425754e-06, "epoch": 1.96, "step": 52970}
+{"loss": 0.5864, "learning_rate": 5.592068619806171e-06, "epoch": 1.96, "step": 52980}
+{"loss": 0.5436, "learning_rate": 5.536370725186588e-06, "epoch": 1.96, "step": 52990}
+{"loss": 0.5825, "learning_rate": 5.486242620028963e-06, "epoch": 1.96, "step": 53000}
+{"loss": 0.5749, "learning_rate": 5.430544725409379e-06, "epoch": 1.96, "step": 53010}
+{"loss": 0.5666, "learning_rate": 5.374846830789796e-06, "epoch": 1.97, "step": 53020}
+{"loss": 0.5751, "learning_rate": 5.319148936170213e-06, "epoch": 1.97, "step": 53030}
+{"loss": 0.7009, "learning_rate": 5.263451041550629e-06, "epoch": 1.97, "step": 53040}
+{"loss": 0.6883, "learning_rate": 5.207753146931045e-06, "epoch": 1.97, "step": 53050}
+{"loss": 0.5404, "learning_rate": 5.152055252311462e-06, "epoch": 1.97, "step": 53060}
+{"loss": 0.4765, "learning_rate": 5.096357357691879e-06, "epoch": 1.97, "step": 53070}
+{"loss": 0.4525, "learning_rate": 5.040659463072295e-06, "epoch": 1.97, "step": 53080}
+{"loss": 0.5388, "learning_rate": 4.984961568452712e-06, "epoch": 1.97, "step": 53090}
+{"loss": 0.43, "learning_rate": 4.929263673833128e-06, "epoch": 1.97, "step": 53100}
+{"loss": 0.5274, "learning_rate": 4.873565779213545e-06, "epoch": 1.97, "step": 53110}
+{"loss": 0.5246, "learning_rate": 4.8178678845939616e-06, "epoch": 1.97, "step": 53120}
+{"loss": 0.5637, "learning_rate": 4.762169989974379e-06, "epoch": 1.97, "step": 53130}
+{"loss": 0.5562, "learning_rate": 4.706472095354795e-06, "epoch": 1.97, "step": 53140}
+{"loss": 0.582, "learning_rate": 4.650774200735212e-06, "epoch": 1.97, "step": 53150}
+{"loss": 0.5761, "learning_rate": 4.595076306115629e-06, "epoch": 1.97, "step": 53160}
+{"loss": 0.7622, "learning_rate": 4.539378411496045e-06, "epoch": 1.97, "step": 53170}
+{"loss": 0.5229, "learning_rate": 4.483680516876461e-06, "epoch": 1.97, "step": 53180}
+{"loss": 0.5456, "learning_rate": 4.4279826222568785e-06, "epoch": 1.97, "step": 53190}
+{"loss": 0.5782, "learning_rate": 4.372284727637295e-06, "epoch": 1.97, "step": 53200}
+{"loss": 0.6034, "learning_rate": 4.316586833017712e-06, "epoch": 1.97, "step": 53210}
+{"loss": 0.369, "learning_rate": 4.260888938398128e-06, "epoch": 1.97, "step": 53220}
+{"loss": 0.6394, "learning_rate": 4.205191043778545e-06, "epoch": 1.97, "step": 53230}
+{"loss": 0.6187, "learning_rate": 4.149493149158961e-06, "epoch": 1.97, "step": 53240}
+{"loss": 0.5428, "learning_rate": 4.0937952545393775e-06, "epoch": 1.97, "step": 53250}
+{"loss": 0.6027, "learning_rate": 4.0380973599197945e-06, "epoch": 1.97, "step": 53260}
+{"loss": 0.533, "learning_rate": 3.982399465300212e-06, "epoch": 1.97, "step": 53270}
+{"loss": 0.4162, "learning_rate": 3.926701570680628e-06, "epoch": 1.97, "step": 53280}
+{"loss": 0.5177, "learning_rate": 3.871003676061045e-06, "epoch": 1.98, "step": 53290}
+{"loss": 0.5224, "learning_rate": 3.815305781441461e-06, "epoch": 1.98, "step": 53300}
+{"loss": 0.6041, "learning_rate": 3.759607886821878e-06, "epoch": 1.98, "step": 53310}
+{"loss": 0.5063, "learning_rate": 3.703909992202295e-06, "epoch": 1.98, "step": 53320}
+{"loss": 0.6732, "learning_rate": 3.648212097582711e-06, "epoch": 1.98, "step": 53330}
+{"loss": 0.5758, "learning_rate": 3.5925142029631277e-06, "epoch": 1.98, "step": 53340}
+{"loss": 0.6321, "learning_rate": 3.5368163083435443e-06, "epoch": 1.98, "step": 53350}
+{"loss": 0.4584, "learning_rate": 3.481118413723961e-06, "epoch": 1.98, "step": 53360}
+{"loss": 0.5924, "learning_rate": 3.4254205191043776e-06, "epoch": 1.98, "step": 53370}
+{"loss": 0.5927, "learning_rate": 3.3697226244847942e-06, "epoch": 1.98, "step": 53380}
+{"loss": 0.594, "learning_rate": 3.314024729865211e-06, "epoch": 1.98, "step": 53390}
+{"loss": 0.5333, "learning_rate": 3.2583268352456275e-06, "epoch": 1.98, "step": 53400}
+{"loss": 0.5619, "learning_rate": 3.2026289406260437e-06, "epoch": 1.98, "step": 53410}
+{"loss": 0.742, "learning_rate": 3.1469310460064604e-06, "epoch": 1.98, "step": 53420}
+{"loss": 0.5809, "learning_rate": 3.0912331513868775e-06, "epoch": 1.98, "step": 53430}
+{"loss": 0.4999, "learning_rate": 3.035535256767294e-06, "epoch": 1.98, "step": 53440}
+{"loss": 0.5197, "learning_rate": 2.9798373621477107e-06, "epoch": 1.98, "step": 53450}
+{"loss": 0.6381, "learning_rate": 2.9241394675281274e-06, "epoch": 1.98, "step": 53460}
+{"loss": 0.706, "learning_rate": 2.868441572908544e-06, "epoch": 1.98, "step": 53470}
+{"loss": 0.483, "learning_rate": 2.8127436782889602e-06, "epoch": 1.98, "step": 53480}
+{"loss": 0.6196, "learning_rate": 2.757045783669377e-06, "epoch": 1.98, "step": 53490}
+{"loss": 0.542, "learning_rate": 2.7013478890497935e-06, "epoch": 1.98, "step": 53500}
+{"loss": 0.603, "learning_rate": 2.64564999443021e-06, "epoch": 1.98, "step": 53510}
+{"loss": 0.6736, "learning_rate": 2.589952099810627e-06, "epoch": 1.98, "step": 53520}
+{"loss": 0.4472, "learning_rate": 2.5342542051910434e-06, "epoch": 1.98, "step": 53530}
+{"loss": 0.6326, "learning_rate": 2.4785563105714605e-06, "epoch": 1.98, "step": 53540}
+{"loss": 0.4913, "learning_rate": 2.422858415951877e-06, "epoch": 1.98, "step": 53550}
+{"loss": 0.587, "learning_rate": 2.3671605213322934e-06, "epoch": 1.99, "step": 53560}
+{"loss": 0.4658, "learning_rate": 2.31146262671271e-06, "epoch": 1.99, "step": 53570}
+{"loss": 0.5464, "learning_rate": 2.2557647320931267e-06, "epoch": 1.99, "step": 53580}
+{"loss": 0.5728, "learning_rate": 2.2000668374735433e-06, "epoch": 1.99, "step": 53590}
+{"loss": 0.5331, "learning_rate": 2.14436894285396e-06, "epoch": 1.99, "step": 53600}
+{"loss": 0.5036, "learning_rate": 2.0886710482343766e-06, "epoch": 1.99, "step": 53610}
+{"loss": 0.4522, "learning_rate": 2.0329731536147932e-06, "epoch": 1.99, "step": 53620}
+{"loss": 0.4594, "learning_rate": 1.97727525899521e-06, "epoch": 1.99, "step": 53630}
+{"loss": 0.5598, "learning_rate": 1.9215773643756265e-06, "epoch": 1.99, "step": 53640}
+{"loss": 0.6121, "learning_rate": 1.8658794697560431e-06, "epoch": 1.99, "step": 53650}
+{"loss": 0.6722, "learning_rate": 1.8101815751364596e-06, "epoch": 1.99, "step": 53660}
+{"loss": 0.4486, "learning_rate": 1.7544836805168762e-06, "epoch": 1.99, "step": 53670}
+{"loss": 0.6013, "learning_rate": 1.698785785897293e-06, "epoch": 1.99, "step": 53680}
+{"loss": 0.4943, "learning_rate": 1.6430878912777097e-06, "epoch": 1.99, "step": 53690}
+{"loss": 0.6126, "learning_rate": 1.5873899966581261e-06, "epoch": 1.99, "step": 53700}
+{"loss": 0.5886, "learning_rate": 1.5316921020385428e-06, "epoch": 1.99, "step": 53710}
+{"loss": 0.5682, "learning_rate": 1.4759942074189594e-06, "epoch": 1.99, "step": 53720}
+{"loss": 0.5473, "learning_rate": 1.4202963127993763e-06, "epoch": 1.99, "step": 53730}
+{"loss": 0.4789, "learning_rate": 1.3645984181797927e-06, "epoch": 1.99, "step": 53740}
+{"loss": 0.4944, "learning_rate": 1.3089005235602093e-06, "epoch": 1.99, "step": 53750}
+{"loss": 0.5667, "learning_rate": 1.253202628940626e-06, "epoch": 1.99, "step": 53760}
+{"loss": 0.5652, "learning_rate": 1.1975047343210426e-06, "epoch": 1.99, "step": 53770}
+{"loss": 0.5463, "learning_rate": 1.1418068397014593e-06, "epoch": 1.99, "step": 53780}
+{"loss": 0.4633, "learning_rate": 1.086108945081876e-06, "epoch": 1.99, "step": 53790}
+{"loss": 0.6169, "learning_rate": 1.0304110504622923e-06, "epoch": 1.99, "step": 53800}
+{"loss": 0.6536, "learning_rate": 9.74713155842709e-07, "epoch": 1.99, "step": 53810}
+{"loss": 0.6189, "learning_rate": 9.190152612231256e-07, "epoch": 1.99, "step": 53820}
+{"loss": 0.4844, "learning_rate": 8.633173666035423e-07, "epoch": 2.0, "step": 53830}
+{"loss": 0.5403, "learning_rate": 8.076194719839589e-07, "epoch": 2.0, "step": 53840}
+{"loss": 0.4962, "learning_rate": 7.519215773643756e-07, "epoch": 2.0, "step": 53850}
+{"loss": 0.4902, "learning_rate": 6.962236827447921e-07, "epoch": 2.0, "step": 53860}
+{"loss": 0.6043, "learning_rate": 6.405257881252088e-07, "epoch": 2.0, "step": 53870}
+{"loss": 0.6135, "learning_rate": 5.848278935056255e-07, "epoch": 2.0, "step": 53880}
+{"loss": 0.6126, "learning_rate": 5.291299988860421e-07, "epoch": 2.0, "step": 53890}
+{"loss": 0.5229, "learning_rate": 4.734321042664587e-07, "epoch": 2.0, "step": 53900}
+{"loss": 0.5079, "learning_rate": 4.1773420964687535e-07, "epoch": 2.0, "step": 53910}
+{"loss": 0.5567, "learning_rate": 3.6203631502729194e-07, "epoch": 2.0, "step": 53920}
+{"loss": 0.5516, "learning_rate": 3.063384204077086e-07, "epoch": 2.0, "step": 53930}
+{"loss": 0.5399, "learning_rate": 2.5064052578812517e-07, "epoch": 2.0, "step": 53940}
+{"loss": 0.4823, "learning_rate": 1.949426311685418e-07, "epoch": 2.0, "step": 53950}
+{"loss": 0.5865, "learning_rate": 1.3924473654895842e-07, "epoch": 2.0, "step": 53960}
+{"train_runtime": 6417.4177, "train_samples_per_second": 8.409, "train_steps_per_second": 8.409, "total_flos": 1.47566567210496e+17, "train_loss": 0.6364423901274552, "epoch": 2.0, "step": 53962}
\ No newline at end of file