diff --git "a/checkpoint-18000/trainer_state.json" "b/checkpoint-18000/trainer_state.json" --- "a/checkpoint-18000/trainer_state.json" +++ "b/checkpoint-18000/trainer_state.json" @@ -1,7 +1,7 @@ { - "best_metric": 0.039903707802295685, - "best_model_checkpoint": "wav2vec2-base-pemlsb-la/checkpoint-18000", - "epoch": 0.9989178390077416, + "best_metric": 0.02988557517528534, + "best_model_checkpoint": "wav2vec2-base-pemlsb-la/checkpoint-14000", + "epoch": 1.9979189211687338, "global_step": 18000, "is_hyper_param_search": false, "is_local_process_zero": true, @@ -10,21688 +10,21688 @@ { "epoch": 0.0, "learning_rate": 1e-05, - "loss": 0.136, + "loss": 0.08, "step": 5 }, { "epoch": 0.0, "learning_rate": 2e-05, - "loss": 0.1461, + "loss": 0.0989, "step": 10 }, { "epoch": 0.0, - "learning_rate": 1.999444722083403e-05, - "loss": 0.1502, + "learning_rate": 1.9994446912483344e-05, + "loss": 0.0956, "step": 15 }, { "epoch": 0.0, - "learning_rate": 1.9988894441668056e-05, - "loss": 0.176, + "learning_rate": 1.9988893824966684e-05, + "loss": 0.1378, "step": 20 }, { "epoch": 0.0, - "learning_rate": 1.9983341662502084e-05, - "loss": 0.1467, + "learning_rate": 1.9983340737450023e-05, + "loss": 0.0434, "step": 25 }, { "epoch": 0.0, - "learning_rate": 1.997778888333611e-05, - "loss": 0.0714, + "learning_rate": 1.9977787649933366e-05, + "loss": 0.1188, "step": 30 }, { "epoch": 0.0, - "learning_rate": 1.9972236104170138e-05, - "loss": 0.1599, + "learning_rate": 1.9972234562416705e-05, + "loss": 0.1167, "step": 35 }, { "epoch": 0.0, - "learning_rate": 1.9966683325004167e-05, - "loss": 0.1974, + "learning_rate": 1.9966681474900048e-05, + "loss": 0.1105, "step": 40 }, { "epoch": 0.0, - "learning_rate": 1.9961130545838195e-05, - "loss": 0.338, + "learning_rate": 1.9961128387383387e-05, + "loss": 0.1941, "step": 45 }, { - "epoch": 0.0, - "learning_rate": 1.995557776667222e-05, - "loss": 0.3956, + "epoch": 0.01, + "learning_rate": 1.9955575299866727e-05, + "loss": 0.2953, "step": 50 }, { - "epoch": 0.0, - "learning_rate": 1.995002498750625e-05, - "loss": 0.1458, + "epoch": 0.01, + "learning_rate": 1.995002221235007e-05, + "loss": 0.1396, "step": 55 }, { - "epoch": 0.0, - "learning_rate": 1.9944472208340275e-05, - "loss": 0.1317, + "epoch": 0.01, + "learning_rate": 1.994446912483341e-05, + "loss": 0.0832, "step": 60 }, { - "epoch": 0.0, - "learning_rate": 1.9938919429174303e-05, - "loss": 0.1742, + "epoch": 0.01, + "learning_rate": 1.9938916037316748e-05, + "loss": 0.1062, "step": 65 }, { - "epoch": 0.0, - "learning_rate": 1.993336665000833e-05, - "loss": 0.1618, + "epoch": 0.01, + "learning_rate": 1.993336294980009e-05, + "loss": 0.1134, "step": 70 }, { - "epoch": 0.0, - "learning_rate": 1.9927813870842357e-05, - "loss": 0.0754, + "epoch": 0.01, + "learning_rate": 1.992780986228343e-05, + "loss": 0.0965, "step": 75 }, { - "epoch": 0.0, - "learning_rate": 1.9922261091676386e-05, - "loss": 0.1022, + "epoch": 0.01, + "learning_rate": 1.9922256774766773e-05, + "loss": 0.0549, "step": 80 }, { - "epoch": 0.0, - "learning_rate": 1.9916708312510414e-05, - "loss": 0.15, + "epoch": 0.01, + "learning_rate": 1.9916703687250112e-05, + "loss": 0.0813, "step": 85 }, { - "epoch": 0.0, - "learning_rate": 1.991115553334444e-05, - "loss": 0.2281, + "epoch": 0.01, + "learning_rate": 1.9911150599733452e-05, + "loss": 0.1101, "step": 90 }, { "epoch": 0.01, - "learning_rate": 1.9905602754178468e-05, - "loss": 0.2028, + "learning_rate": 1.9905597512216795e-05, + "loss": 0.202, "step": 95 }, { "epoch": 0.01, - "learning_rate": 1.9900049975012497e-05, - "loss": 0.3737, + "learning_rate": 1.9900044424700134e-05, + "loss": 0.367, "step": 100 }, { "epoch": 0.01, - "learning_rate": 1.9894497195846522e-05, - "loss": 0.1552, + "learning_rate": 1.9894491337183473e-05, + "loss": 0.1122, "step": 105 }, { "epoch": 0.01, - "learning_rate": 1.988894441668055e-05, - "loss": 0.1664, + "learning_rate": 1.9888938249666816e-05, + "loss": 0.1137, "step": 110 }, { "epoch": 0.01, - "learning_rate": 1.9883391637514576e-05, - "loss": 0.1762, + "learning_rate": 1.988338516215016e-05, + "loss": 0.0945, "step": 115 }, { "epoch": 0.01, - "learning_rate": 1.9877838858348605e-05, - "loss": 0.1818, + "learning_rate": 1.9877832074633498e-05, + "loss": 0.1107, "step": 120 }, { "epoch": 0.01, - "learning_rate": 1.9872286079182633e-05, - "loss": 0.0745, + "learning_rate": 1.9872278987116838e-05, + "loss": 0.0774, "step": 125 }, { "epoch": 0.01, - "learning_rate": 1.9866733300016662e-05, - "loss": 0.1602, + "learning_rate": 1.986672589960018e-05, + "loss": 0.0625, "step": 130 }, { "epoch": 0.01, - "learning_rate": 1.9861180520850687e-05, - "loss": 0.0846, + "learning_rate": 1.986117281208352e-05, + "loss": 0.0663, "step": 135 }, { - "epoch": 0.01, - "learning_rate": 1.9855627741684716e-05, - "loss": 0.2068, + "epoch": 0.02, + "learning_rate": 1.9855619724566862e-05, + "loss": 0.1929, "step": 140 }, { - "epoch": 0.01, - "learning_rate": 1.985007496251874e-05, - "loss": 0.2634, + "epoch": 0.02, + "learning_rate": 1.9850066637050202e-05, + "loss": 0.1206, "step": 145 }, { - "epoch": 0.01, - "learning_rate": 1.984452218335277e-05, - "loss": 0.3582, + "epoch": 0.02, + "learning_rate": 1.984451354953354e-05, + "loss": 0.433, "step": 150 }, { - "epoch": 0.01, - "learning_rate": 1.9838969404186795e-05, - "loss": 0.1664, + "epoch": 0.02, + "learning_rate": 1.9838960462016884e-05, + "loss": 0.103, "step": 155 }, { - "epoch": 0.01, - "learning_rate": 1.9833416625020824e-05, - "loss": 0.1578, + "epoch": 0.02, + "learning_rate": 1.9833407374500223e-05, + "loss": 0.0935, "step": 160 }, { - "epoch": 0.01, - "learning_rate": 1.9827863845854852e-05, - "loss": 0.1828, + "epoch": 0.02, + "learning_rate": 1.9827854286983566e-05, + "loss": 0.1095, "step": 165 }, { - "epoch": 0.01, - "learning_rate": 1.982231106668888e-05, - "loss": 0.1735, + "epoch": 0.02, + "learning_rate": 1.9822301199466905e-05, + "loss": 0.1218, "step": 170 }, { - "epoch": 0.01, - "learning_rate": 1.9816758287522906e-05, - "loss": 0.1939, + "epoch": 0.02, + "learning_rate": 1.9816748111950245e-05, + "loss": 0.0888, "step": 175 }, { - "epoch": 0.01, - "learning_rate": 1.9811205508356935e-05, - "loss": 0.1213, + "epoch": 0.02, + "learning_rate": 1.9811195024433588e-05, + "loss": 0.0929, "step": 180 }, { - "epoch": 0.01, - "learning_rate": 1.980565272919096e-05, - "loss": 0.1186, + "epoch": 0.02, + "learning_rate": 1.9805641936916927e-05, + "loss": 0.1013, "step": 185 }, { - "epoch": 0.01, - "learning_rate": 1.980009995002499e-05, - "loss": 0.1388, + "epoch": 0.02, + "learning_rate": 1.9800088849400266e-05, + "loss": 0.136, "step": 190 }, { - "epoch": 0.01, - "learning_rate": 1.9794547170859017e-05, - "loss": 0.2075, + "epoch": 0.02, + "learning_rate": 1.979453576188361e-05, + "loss": 0.1258, "step": 195 }, { - "epoch": 0.01, - "learning_rate": 1.9788994391693043e-05, - "loss": 0.4645, + "epoch": 0.02, + "learning_rate": 1.978898267436695e-05, + "loss": 0.3627, "step": 200 }, { - "epoch": 0.01, - "learning_rate": 1.978344161252707e-05, - "loss": 0.1823, + "epoch": 0.02, + "learning_rate": 1.978342958685029e-05, + "loss": 0.0882, "step": 205 }, { - "epoch": 0.01, - "learning_rate": 1.97778888333611e-05, - "loss": 0.1642, + "epoch": 0.02, + "learning_rate": 1.977787649933363e-05, + "loss": 0.1098, "step": 210 }, { - "epoch": 0.01, - "learning_rate": 1.977233605419513e-05, - "loss": 0.1501, + "epoch": 0.02, + "learning_rate": 1.977232341181697e-05, + "loss": 0.0994, "step": 215 }, { - "epoch": 0.01, - "learning_rate": 1.9766783275029154e-05, - "loss": 0.1343, + "epoch": 0.02, + "learning_rate": 1.9766770324300313e-05, + "loss": 0.1581, "step": 220 }, { - "epoch": 0.01, - "learning_rate": 1.9761230495863182e-05, - "loss": 0.1088, + "epoch": 0.02, + "learning_rate": 1.9761217236783655e-05, + "loss": 0.0698, "step": 225 }, { - "epoch": 0.01, - "learning_rate": 1.9755677716697208e-05, - "loss": 0.0794, + "epoch": 0.03, + "learning_rate": 1.9755664149266995e-05, + "loss": 0.1023, "step": 230 }, { - "epoch": 0.01, - "learning_rate": 1.9750124937531236e-05, - "loss": 0.0869, + "epoch": 0.03, + "learning_rate": 1.9750111061750334e-05, + "loss": 0.0995, "step": 235 }, { - "epoch": 0.01, - "learning_rate": 1.974457215836526e-05, - "loss": 0.194, + "epoch": 0.03, + "learning_rate": 1.9744557974233677e-05, + "loss": 0.1318, "step": 240 }, { - "epoch": 0.01, - "learning_rate": 1.973901937919929e-05, - "loss": 0.3068, + "epoch": 0.03, + "learning_rate": 1.9739004886717016e-05, + "loss": 0.2402, "step": 245 }, { - "epoch": 0.01, - "learning_rate": 1.973346660003332e-05, - "loss": 0.5181, + "epoch": 0.03, + "learning_rate": 1.9733451799200356e-05, + "loss": 0.4634, "step": 250 }, { - "epoch": 0.01, - "learning_rate": 1.9727913820867348e-05, - "loss": 0.1538, + "epoch": 0.03, + "learning_rate": 1.97278987116837e-05, + "loss": 0.0771, "step": 255 }, { - "epoch": 0.01, - "learning_rate": 1.9722361041701373e-05, - "loss": 0.149, + "epoch": 0.03, + "learning_rate": 1.9722345624167038e-05, + "loss": 0.0877, "step": 260 }, { - "epoch": 0.01, - "learning_rate": 1.97168082625354e-05, - "loss": 0.1142, + "epoch": 0.03, + "learning_rate": 1.971679253665038e-05, + "loss": 0.1057, "step": 265 }, { - "epoch": 0.01, - "learning_rate": 1.9711255483369427e-05, - "loss": 0.2104, + "epoch": 0.03, + "learning_rate": 1.971123944913372e-05, + "loss": 0.094, "step": 270 }, { - "epoch": 0.02, - "learning_rate": 1.9705702704203455e-05, - "loss": 0.1418, + "epoch": 0.03, + "learning_rate": 1.970568636161706e-05, + "loss": 0.0814, "step": 275 }, { - "epoch": 0.02, - "learning_rate": 1.9700149925037484e-05, - "loss": 0.073, + "epoch": 0.03, + "learning_rate": 1.9700133274100402e-05, + "loss": 0.0621, "step": 280 }, { - "epoch": 0.02, - "learning_rate": 1.969459714587151e-05, - "loss": 0.0998, + "epoch": 0.03, + "learning_rate": 1.969458018658374e-05, + "loss": 0.111, "step": 285 }, { - "epoch": 0.02, - "learning_rate": 1.9689044366705538e-05, - "loss": 0.1924, + "epoch": 0.03, + "learning_rate": 1.9689027099067084e-05, + "loss": 0.1037, "step": 290 }, { - "epoch": 0.02, - "learning_rate": 1.968460214337276e-05, - "loss": 0.2824, + "epoch": 0.03, + "learning_rate": 1.9683474011550424e-05, + "loss": 0.1733, "step": 295 }, { - "epoch": 0.02, - "learning_rate": 1.967904936420679e-05, - "loss": 0.4568, + "epoch": 0.03, + "learning_rate": 1.9677920924033763e-05, + "loss": 0.3142, "step": 300 }, { - "epoch": 0.02, - "learning_rate": 1.9673496585040815e-05, - "loss": 0.1453, + "epoch": 0.03, + "learning_rate": 1.9672367836517106e-05, + "loss": 0.0859, "step": 305 }, { - "epoch": 0.02, - "learning_rate": 1.9667943805874843e-05, - "loss": 0.1231, + "epoch": 0.03, + "learning_rate": 1.966681474900045e-05, + "loss": 0.1049, "step": 310 }, { - "epoch": 0.02, - "learning_rate": 1.966239102670887e-05, - "loss": 0.2111, + "epoch": 0.03, + "learning_rate": 1.9661261661483784e-05, + "loss": 0.1117, "step": 315 }, { - "epoch": 0.02, - "learning_rate": 1.9656838247542897e-05, - "loss": 0.1434, + "epoch": 0.04, + "learning_rate": 1.9655708573967127e-05, + "loss": 0.1147, "step": 320 }, { - "epoch": 0.02, - "learning_rate": 1.9651285468376922e-05, - "loss": 0.1642, + "epoch": 0.04, + "learning_rate": 1.965015548645047e-05, + "loss": 0.0874, "step": 325 }, { - "epoch": 0.02, - "learning_rate": 1.964573268921095e-05, - "loss": 0.1127, + "epoch": 0.04, + "learning_rate": 1.964460239893381e-05, + "loss": 0.1098, "step": 330 }, { - "epoch": 0.02, - "learning_rate": 1.964017991004498e-05, - "loss": 0.1189, + "epoch": 0.04, + "learning_rate": 1.963904931141715e-05, + "loss": 0.085, "step": 335 }, { - "epoch": 0.02, - "learning_rate": 1.963462713087901e-05, - "loss": 0.1372, + "epoch": 0.04, + "learning_rate": 1.963349622390049e-05, + "loss": 0.1585, "step": 340 }, { - "epoch": 0.02, - "learning_rate": 1.9629074351713034e-05, - "loss": 0.2152, + "epoch": 0.04, + "learning_rate": 1.962794313638383e-05, + "loss": 0.2213, "step": 345 }, { - "epoch": 0.02, - "learning_rate": 1.9623521572547062e-05, - "loss": 0.3897, + "epoch": 0.04, + "learning_rate": 1.9622390048867174e-05, + "loss": 0.4533, "step": 350 }, { - "epoch": 0.02, - "learning_rate": 1.9617968793381088e-05, - "loss": 0.1409, + "epoch": 0.04, + "learning_rate": 1.9616836961350513e-05, + "loss": 0.1004, "step": 355 }, { - "epoch": 0.02, - "learning_rate": 1.9612416014215116e-05, - "loss": 0.0985, + "epoch": 0.04, + "learning_rate": 1.9611283873833852e-05, + "loss": 0.0968, "step": 360 }, { - "epoch": 0.02, - "learning_rate": 1.960686323504914e-05, - "loss": 0.1778, + "epoch": 0.04, + "learning_rate": 1.9605730786317195e-05, + "loss": 0.1071, "step": 365 }, { - "epoch": 0.02, - "learning_rate": 1.960131045588317e-05, - "loss": 0.153, + "epoch": 0.04, + "learning_rate": 1.9600177698800534e-05, + "loss": 0.1426, "step": 370 }, { - "epoch": 0.02, - "learning_rate": 1.95957576767172e-05, - "loss": 0.1419, + "epoch": 0.04, + "learning_rate": 1.9594624611283874e-05, + "loss": 0.1237, "step": 375 }, { - "epoch": 0.02, - "learning_rate": 1.9590204897551227e-05, - "loss": 0.1058, + "epoch": 0.04, + "learning_rate": 1.9589071523767217e-05, + "loss": 0.0596, "step": 380 }, { - "epoch": 0.02, - "learning_rate": 1.9584652118385256e-05, - "loss": 0.1321, + "epoch": 0.04, + "learning_rate": 1.9583518436250556e-05, + "loss": 0.0578, "step": 385 }, { - "epoch": 0.02, - "learning_rate": 1.957909933921928e-05, - "loss": 0.2512, + "epoch": 0.04, + "learning_rate": 1.95779653487339e-05, + "loss": 0.1257, "step": 390 }, { - "epoch": 0.02, - "learning_rate": 1.957354656005331e-05, - "loss": 0.1711, + "epoch": 0.04, + "learning_rate": 1.9572412261217238e-05, + "loss": 0.1492, "step": 395 }, { - "epoch": 0.02, - "learning_rate": 1.9567993780887335e-05, - "loss": 0.3087, + "epoch": 0.04, + "learning_rate": 1.9566859173700577e-05, + "loss": 0.2997, "step": 400 }, { - "epoch": 0.02, - "learning_rate": 1.9562441001721364e-05, - "loss": 0.1076, + "epoch": 0.04, + "learning_rate": 1.956130608618392e-05, + "loss": 0.1145, "step": 405 }, { - "epoch": 0.02, - "learning_rate": 1.955688822255539e-05, - "loss": 0.1542, + "epoch": 0.05, + "learning_rate": 1.955575299866726e-05, + "loss": 0.1076, "step": 410 }, { - "epoch": 0.02, - "learning_rate": 1.9551335443389418e-05, - "loss": 0.1514, + "epoch": 0.05, + "learning_rate": 1.95501999111506e-05, + "loss": 0.0894, "step": 415 }, { - "epoch": 0.02, - "learning_rate": 1.9545782664223446e-05, - "loss": 0.1578, + "epoch": 0.05, + "learning_rate": 1.954464682363394e-05, + "loss": 0.1168, "step": 420 }, { - "epoch": 0.02, - "learning_rate": 1.9540229885057475e-05, - "loss": 0.0895, + "epoch": 0.05, + "learning_rate": 1.9539093736117284e-05, + "loss": 0.0885, "step": 425 }, { - "epoch": 0.02, - "learning_rate": 1.95346771058915e-05, - "loss": 0.097, + "epoch": 0.05, + "learning_rate": 1.9533540648600624e-05, + "loss": 0.0653, "step": 430 }, { - "epoch": 0.02, - "learning_rate": 1.952912432672553e-05, - "loss": 0.1601, + "epoch": 0.05, + "learning_rate": 1.9527987561083967e-05, + "loss": 0.0617, "step": 435 }, { - "epoch": 0.02, - "learning_rate": 1.9523571547559554e-05, - "loss": 0.2003, + "epoch": 0.05, + "learning_rate": 1.9522434473567306e-05, + "loss": 0.1256, "step": 440 }, { - "epoch": 0.02, - "learning_rate": 1.9518018768393583e-05, - "loss": 0.2784, + "epoch": 0.05, + "learning_rate": 1.9516881386050645e-05, + "loss": 0.238, "step": 445 }, { - "epoch": 0.02, - "learning_rate": 1.9512465989227608e-05, - "loss": 0.5125, + "epoch": 0.05, + "learning_rate": 1.9511328298533988e-05, + "loss": 0.326, "step": 450 }, { - "epoch": 0.03, - "learning_rate": 1.9506913210061637e-05, - "loss": 0.1948, + "epoch": 0.05, + "learning_rate": 1.9505775211017327e-05, + "loss": 0.118, "step": 455 }, { - "epoch": 0.03, - "learning_rate": 1.9501360430895665e-05, - "loss": 0.1439, + "epoch": 0.05, + "learning_rate": 1.9500222123500667e-05, + "loss": 0.0939, "step": 460 }, { - "epoch": 0.03, - "learning_rate": 1.9495807651729694e-05, - "loss": 0.1542, + "epoch": 0.05, + "learning_rate": 1.949466903598401e-05, + "loss": 0.121, "step": 465 }, { - "epoch": 0.03, - "learning_rate": 1.949025487256372e-05, - "loss": 0.1546, + "epoch": 0.05, + "learning_rate": 1.948911594846735e-05, + "loss": 0.1211, "step": 470 }, { - "epoch": 0.03, - "learning_rate": 1.9484702093397748e-05, - "loss": 0.1167, + "epoch": 0.05, + "learning_rate": 1.948356286095069e-05, + "loss": 0.0962, "step": 475 }, { - "epoch": 0.03, - "learning_rate": 1.9479149314231777e-05, - "loss": 0.0682, + "epoch": 0.05, + "learning_rate": 1.947800977343403e-05, + "loss": 0.059, "step": 480 }, { - "epoch": 0.03, - "learning_rate": 1.9473596535065802e-05, - "loss": 0.1299, + "epoch": 0.05, + "learning_rate": 1.947245668591737e-05, + "loss": 0.1039, "step": 485 }, { - "epoch": 0.03, - "learning_rate": 1.946804375589983e-05, - "loss": 0.1928, + "epoch": 0.05, + "learning_rate": 1.9468014215904045e-05, + "loss": 0.1474, "step": 490 }, { - "epoch": 0.03, - "learning_rate": 1.9462490976733856e-05, - "loss": 0.1997, + "epoch": 0.05, + "learning_rate": 1.9462461128387385e-05, + "loss": 0.1206, "step": 495 }, { - "epoch": 0.03, - "learning_rate": 1.9456938197567884e-05, - "loss": 0.5177, + "epoch": 0.06, + "learning_rate": 1.9456908040870724e-05, + "loss": 0.4527, "step": 500 }, { - "epoch": 0.03, - "learning_rate": 1.9451385418401913e-05, - "loss": 0.2133, + "epoch": 0.06, + "learning_rate": 1.9451354953354067e-05, + "loss": 0.1067, "step": 505 }, { - "epoch": 0.03, - "learning_rate": 1.9445832639235938e-05, - "loss": 0.1063, + "epoch": 0.06, + "learning_rate": 1.944580186583741e-05, + "loss": 0.094, "step": 510 }, { - "epoch": 0.03, - "learning_rate": 1.9440279860069967e-05, - "loss": 0.1528, + "epoch": 0.06, + "learning_rate": 1.944024877832075e-05, + "loss": 0.0877, "step": 515 }, { - "epoch": 0.03, - "learning_rate": 1.9434727080903996e-05, - "loss": 0.1154, + "epoch": 0.06, + "learning_rate": 1.9434695690804088e-05, + "loss": 0.0908, "step": 520 }, { - "epoch": 0.03, - "learning_rate": 1.942917430173802e-05, - "loss": 0.0912, + "epoch": 0.06, + "learning_rate": 1.942914260328743e-05, + "loss": 0.0999, "step": 525 }, { - "epoch": 0.03, - "learning_rate": 1.942362152257205e-05, - "loss": 0.1487, + "epoch": 0.06, + "learning_rate": 1.942358951577077e-05, + "loss": 0.0366, "step": 530 }, { - "epoch": 0.03, - "learning_rate": 1.9418068743406075e-05, - "loss": 0.1682, + "epoch": 0.06, + "learning_rate": 1.941803642825411e-05, + "loss": 0.0938, "step": 535 }, { - "epoch": 0.03, - "learning_rate": 1.9412515964240103e-05, - "loss": 0.2485, + "epoch": 0.06, + "learning_rate": 1.9412483340737452e-05, + "loss": 0.1235, "step": 540 }, { - "epoch": 0.03, - "learning_rate": 1.940696318507413e-05, - "loss": 0.2143, + "epoch": 0.06, + "learning_rate": 1.9406930253220792e-05, + "loss": 0.1556, "step": 545 }, { - "epoch": 0.03, - "learning_rate": 1.9401410405908157e-05, - "loss": 0.4632, + "epoch": 0.06, + "learning_rate": 1.9401377165704135e-05, + "loss": 0.2944, "step": 550 }, { - "epoch": 0.03, - "learning_rate": 1.9395857626742186e-05, - "loss": 0.097, + "epoch": 0.06, + "learning_rate": 1.9395824078187474e-05, + "loss": 0.0929, "step": 555 }, { - "epoch": 0.03, - "learning_rate": 1.9390304847576214e-05, - "loss": 0.1619, + "epoch": 0.06, + "learning_rate": 1.9390270990670813e-05, + "loss": 0.1046, "step": 560 }, { - "epoch": 0.03, - "learning_rate": 1.9384752068410243e-05, - "loss": 0.1516, + "epoch": 0.06, + "learning_rate": 1.9384717903154156e-05, + "loss": 0.0832, "step": 565 }, { - "epoch": 0.03, - "learning_rate": 1.937919928924427e-05, - "loss": 0.1512, + "epoch": 0.06, + "learning_rate": 1.9379164815637495e-05, + "loss": 0.105, "step": 570 }, { - "epoch": 0.03, - "learning_rate": 1.9373646510078297e-05, - "loss": 0.1128, + "epoch": 0.06, + "learning_rate": 1.9373611728120835e-05, + "loss": 0.1068, "step": 575 }, { - "epoch": 0.03, - "learning_rate": 1.9368093730912322e-05, - "loss": 0.1092, + "epoch": 0.06, + "learning_rate": 1.9368058640604178e-05, + "loss": 0.0657, "step": 580 }, { - "epoch": 0.03, - "learning_rate": 1.936254095174635e-05, - "loss": 0.2087, + "epoch": 0.06, + "learning_rate": 1.9362505553087517e-05, + "loss": 0.1071, "step": 585 }, { - "epoch": 0.03, - "learning_rate": 1.9356988172580376e-05, - "loss": 0.1076, + "epoch": 0.07, + "learning_rate": 1.935695246557086e-05, + "loss": 0.132, "step": 590 }, { - "epoch": 0.03, - "learning_rate": 1.9351435393414405e-05, - "loss": 0.2047, + "epoch": 0.07, + "learning_rate": 1.93513993780542e-05, + "loss": 0.1616, "step": 595 }, { - "epoch": 0.03, - "learning_rate": 1.9345882614248433e-05, - "loss": 0.4113, + "epoch": 0.07, + "learning_rate": 1.934584629053754e-05, + "loss": 0.389, "step": 600 }, { - "epoch": 0.03, - "learning_rate": 1.9340329835082462e-05, - "loss": 0.1408, + "epoch": 0.07, + "learning_rate": 1.934029320302088e-05, + "loss": 0.0847, "step": 605 }, { - "epoch": 0.03, - "learning_rate": 1.9334777055916487e-05, - "loss": 0.2451, + "epoch": 0.07, + "learning_rate": 1.9334740115504224e-05, + "loss": 0.0961, "step": 610 }, { - "epoch": 0.03, - "learning_rate": 1.9329224276750516e-05, - "loss": 0.1125, + "epoch": 0.07, + "learning_rate": 1.9329187027987563e-05, + "loss": 0.0977, "step": 615 }, { - "epoch": 0.03, - "learning_rate": 1.932367149758454e-05, - "loss": 0.1436, + "epoch": 0.07, + "learning_rate": 1.9323633940470903e-05, + "loss": 0.1434, "step": 620 }, { - "epoch": 0.03, - "learning_rate": 1.931811871841857e-05, - "loss": 0.0601, + "epoch": 0.07, + "learning_rate": 1.9318080852954245e-05, + "loss": 0.0628, "step": 625 }, { - "epoch": 0.03, - "learning_rate": 1.9312565939252595e-05, - "loss": 0.1019, + "epoch": 0.07, + "learning_rate": 1.9312527765437585e-05, + "loss": 0.1, "step": 630 }, { - "epoch": 0.04, - "learning_rate": 1.9307013160086624e-05, - "loss": 0.1571, + "epoch": 0.07, + "learning_rate": 1.9306974677920928e-05, + "loss": 0.1099, "step": 635 }, { - "epoch": 0.04, - "learning_rate": 1.9301460380920652e-05, - "loss": 0.2072, + "epoch": 0.07, + "learning_rate": 1.9301421590404267e-05, + "loss": 0.0929, "step": 640 }, { - "epoch": 0.04, - "learning_rate": 1.929590760175468e-05, - "loss": 0.2469, + "epoch": 0.07, + "learning_rate": 1.9295868502887606e-05, + "loss": 0.1874, "step": 645 }, { - "epoch": 0.04, - "learning_rate": 1.929035482258871e-05, - "loss": 0.545, + "epoch": 0.07, + "learning_rate": 1.929031541537095e-05, + "loss": 0.2716, "step": 650 }, { - "epoch": 0.04, - "learning_rate": 1.9284802043422735e-05, - "loss": 0.1186, + "epoch": 0.07, + "learning_rate": 1.928476232785429e-05, + "loss": 0.0824, "step": 655 }, { - "epoch": 0.04, - "learning_rate": 1.9279249264256764e-05, - "loss": 0.1851, + "epoch": 0.07, + "learning_rate": 1.9279209240337628e-05, + "loss": 0.1086, "step": 660 }, { - "epoch": 0.04, - "learning_rate": 1.927369648509079e-05, - "loss": 0.1665, + "epoch": 0.07, + "learning_rate": 1.927365615282097e-05, + "loss": 0.0988, "step": 665 }, { - "epoch": 0.04, - "learning_rate": 1.9268143705924818e-05, - "loss": 0.2188, + "epoch": 0.07, + "learning_rate": 1.926810306530431e-05, + "loss": 0.0907, "step": 670 }, { - "epoch": 0.04, - "learning_rate": 1.9262590926758843e-05, - "loss": 0.1844, + "epoch": 0.07, + "learning_rate": 1.9262549977787653e-05, + "loss": 0.0858, "step": 675 }, { - "epoch": 0.04, - "learning_rate": 1.925703814759287e-05, - "loss": 0.0983, + "epoch": 0.08, + "learning_rate": 1.9256996890270992e-05, + "loss": 0.0548, "step": 680 }, { - "epoch": 0.04, - "learning_rate": 1.92514853684269e-05, - "loss": 0.0912, + "epoch": 0.08, + "learning_rate": 1.925144380275433e-05, + "loss": 0.1034, "step": 685 }, { - "epoch": 0.04, - "learning_rate": 1.924593258926093e-05, - "loss": 0.2082, + "epoch": 0.08, + "learning_rate": 1.9245890715237674e-05, + "loss": 0.0921, "step": 690 }, { - "epoch": 0.04, - "learning_rate": 1.9240379810094954e-05, - "loss": 0.2413, + "epoch": 0.08, + "learning_rate": 1.9240337627721014e-05, + "loss": 0.1742, "step": 695 }, { - "epoch": 0.04, - "learning_rate": 1.9234827030928983e-05, - "loss": 0.5025, + "epoch": 0.08, + "learning_rate": 1.9234784540204353e-05, + "loss": 0.3367, "step": 700 }, { - "epoch": 0.04, - "learning_rate": 1.9229274251763008e-05, - "loss": 0.1503, + "epoch": 0.08, + "learning_rate": 1.9229231452687696e-05, + "loss": 0.0953, "step": 705 }, { - "epoch": 0.04, - "learning_rate": 1.9223721472597036e-05, - "loss": 0.1466, + "epoch": 0.08, + "learning_rate": 1.922367836517104e-05, + "loss": 0.078, "step": 710 }, { - "epoch": 0.04, - "learning_rate": 1.9218168693431062e-05, - "loss": 0.1328, + "epoch": 0.08, + "learning_rate": 1.9218125277654378e-05, + "loss": 0.1181, "step": 715 }, { - "epoch": 0.04, - "learning_rate": 1.921261591426509e-05, - "loss": 0.1838, + "epoch": 0.08, + "learning_rate": 1.9212572190137717e-05, + "loss": 0.0898, "step": 720 }, { - "epoch": 0.04, - "learning_rate": 1.920706313509912e-05, - "loss": 0.1417, + "epoch": 0.08, + "learning_rate": 1.920701910262106e-05, + "loss": 0.0577, "step": 725 }, { - "epoch": 0.04, - "learning_rate": 1.9201510355933148e-05, - "loss": 0.0639, + "epoch": 0.08, + "learning_rate": 1.92014660151044e-05, + "loss": 0.0555, "step": 730 }, { - "epoch": 0.04, - "learning_rate": 1.9195957576767176e-05, - "loss": 0.0972, + "epoch": 0.08, + "learning_rate": 1.9195912927587742e-05, + "loss": 0.1174, "step": 735 }, { - "epoch": 0.04, - "learning_rate": 1.91904047976012e-05, - "loss": 0.1138, + "epoch": 0.08, + "learning_rate": 1.919035984007108e-05, + "loss": 0.1356, "step": 740 }, { - "epoch": 0.04, - "learning_rate": 1.918485201843523e-05, - "loss": 0.2788, + "epoch": 0.08, + "learning_rate": 1.918480675255442e-05, + "loss": 0.1978, "step": 745 }, { - "epoch": 0.04, - "learning_rate": 1.9179299239269255e-05, - "loss": 0.3735, + "epoch": 0.08, + "learning_rate": 1.9179253665037764e-05, + "loss": 0.3164, "step": 750 }, { - "epoch": 0.04, - "learning_rate": 1.9173746460103284e-05, - "loss": 0.1961, + "epoch": 0.08, + "learning_rate": 1.9173700577521103e-05, + "loss": 0.0961, "step": 755 }, { - "epoch": 0.04, - "learning_rate": 1.916819368093731e-05, - "loss": 0.1248, + "epoch": 0.08, + "learning_rate": 1.9168147490004446e-05, + "loss": 0.0904, "step": 760 }, { - "epoch": 0.04, - "learning_rate": 1.9162640901771338e-05, - "loss": 0.1514, + "epoch": 0.08, + "learning_rate": 1.9162594402487785e-05, + "loss": 0.1158, "step": 765 }, { - "epoch": 0.04, - "learning_rate": 1.9157088122605367e-05, - "loss": 0.1685, + "epoch": 0.09, + "learning_rate": 1.9157041314971124e-05, + "loss": 0.1316, "step": 770 }, { - "epoch": 0.04, - "learning_rate": 1.9151535343439392e-05, - "loss": 0.0989, + "epoch": 0.09, + "learning_rate": 1.9151488227454467e-05, + "loss": 0.076, "step": 775 }, { - "epoch": 0.04, - "learning_rate": 1.914598256427342e-05, - "loss": 0.0736, + "epoch": 0.09, + "learning_rate": 1.9145935139937807e-05, + "loss": 0.0589, "step": 780 }, { - "epoch": 0.04, - "learning_rate": 1.914042978510745e-05, - "loss": 0.0646, + "epoch": 0.09, + "learning_rate": 1.9140382052421146e-05, + "loss": 0.0889, "step": 785 }, { - "epoch": 0.04, - "learning_rate": 1.9134877005941474e-05, - "loss": 0.1529, + "epoch": 0.09, + "learning_rate": 1.913482896490449e-05, + "loss": 0.1223, "step": 790 }, { - "epoch": 0.04, - "learning_rate": 1.9129324226775503e-05, - "loss": 0.2095, + "epoch": 0.09, + "learning_rate": 1.9129275877387828e-05, + "loss": 0.1641, "step": 795 }, { - "epoch": 0.04, - "learning_rate": 1.912377144760953e-05, - "loss": 0.4124, + "epoch": 0.09, + "learning_rate": 1.912372278987117e-05, + "loss": 0.3307, "step": 800 }, { - "epoch": 0.04, - "learning_rate": 1.9118218668443557e-05, - "loss": 0.1602, + "epoch": 0.09, + "learning_rate": 1.911816970235451e-05, + "loss": 0.1, "step": 805 }, { - "epoch": 0.04, - "learning_rate": 1.9112665889277586e-05, - "loss": 0.1736, + "epoch": 0.09, + "learning_rate": 1.911261661483785e-05, + "loss": 0.0979, "step": 810 }, { - "epoch": 0.05, - "learning_rate": 1.910711311011161e-05, - "loss": 0.1485, + "epoch": 0.09, + "learning_rate": 1.9107063527321192e-05, + "loss": 0.1066, "step": 815 }, { - "epoch": 0.05, - "learning_rate": 1.910156033094564e-05, - "loss": 0.1431, + "epoch": 0.09, + "learning_rate": 1.9101510439804535e-05, + "loss": 0.1103, "step": 820 }, { - "epoch": 0.05, - "learning_rate": 1.9096007551779668e-05, - "loss": 0.0927, + "epoch": 0.09, + "learning_rate": 1.9095957352287874e-05, + "loss": 0.0711, "step": 825 }, { - "epoch": 0.05, - "learning_rate": 1.9090454772613697e-05, - "loss": 0.1009, + "epoch": 0.09, + "learning_rate": 1.9090404264771214e-05, + "loss": 0.0922, "step": 830 }, { - "epoch": 0.05, - "learning_rate": 1.9084901993447722e-05, - "loss": 0.1008, + "epoch": 0.09, + "learning_rate": 1.9084851177254557e-05, + "loss": 0.0949, "step": 835 }, { - "epoch": 0.05, - "learning_rate": 1.907934921428175e-05, - "loss": 0.1555, + "epoch": 0.09, + "learning_rate": 1.9079298089737896e-05, + "loss": 0.0693, "step": 840 }, { - "epoch": 0.05, - "learning_rate": 1.9073796435115776e-05, - "loss": 0.2275, + "epoch": 0.09, + "learning_rate": 1.9073745002221235e-05, + "loss": 0.1918, "step": 845 }, { - "epoch": 0.05, - "learning_rate": 1.9068243655949805e-05, - "loss": 0.4855, + "epoch": 0.09, + "learning_rate": 1.9068191914704578e-05, + "loss": 0.3019, "step": 850 }, { - "epoch": 0.05, - "learning_rate": 1.906269087678383e-05, - "loss": 0.216, + "epoch": 0.09, + "learning_rate": 1.9062638827187917e-05, + "loss": 0.093, "step": 855 }, { - "epoch": 0.05, - "learning_rate": 1.905713809761786e-05, - "loss": 0.1506, + "epoch": 0.1, + "learning_rate": 1.905708573967126e-05, + "loss": 0.1074, "step": 860 }, { - "epoch": 0.05, - "learning_rate": 1.9051585318451887e-05, - "loss": 0.1468, + "epoch": 0.1, + "learning_rate": 1.90515326521546e-05, + "loss": 0.1087, "step": 865 }, { - "epoch": 0.05, - "learning_rate": 1.9046032539285916e-05, - "loss": 0.142, + "epoch": 0.1, + "learning_rate": 1.904597956463794e-05, + "loss": 0.1158, "step": 870 }, { - "epoch": 0.05, - "learning_rate": 1.904047976011994e-05, - "loss": 0.1425, + "epoch": 0.1, + "learning_rate": 1.904042647712128e-05, + "loss": 0.0637, "step": 875 }, { - "epoch": 0.05, - "learning_rate": 1.903492698095397e-05, - "loss": 0.1158, + "epoch": 0.1, + "learning_rate": 1.903487338960462e-05, + "loss": 0.0768, "step": 880 }, { - "epoch": 0.05, - "learning_rate": 1.9029374201787995e-05, - "loss": 0.1463, + "epoch": 0.1, + "learning_rate": 1.9029320302087964e-05, + "loss": 0.0905, "step": 885 }, { - "epoch": 0.05, - "learning_rate": 1.9023821422622024e-05, - "loss": 0.1914, + "epoch": 0.1, + "learning_rate": 1.9023767214571303e-05, + "loss": 0.0926, "step": 890 }, { - "epoch": 0.05, - "learning_rate": 1.901826864345605e-05, - "loss": 0.3182, + "epoch": 0.1, + "learning_rate": 1.9018214127054642e-05, + "loss": 0.1894, "step": 895 }, { - "epoch": 0.05, - "learning_rate": 1.9012715864290077e-05, - "loss": 0.4656, + "epoch": 0.1, + "learning_rate": 1.9012661039537985e-05, + "loss": 0.412, "step": 900 }, { - "epoch": 0.05, - "learning_rate": 1.9007163085124106e-05, - "loss": 0.1429, + "epoch": 0.1, + "learning_rate": 1.9007107952021328e-05, + "loss": 0.0885, "step": 905 }, { - "epoch": 0.05, - "learning_rate": 1.9001610305958135e-05, - "loss": 0.1415, + "epoch": 0.1, + "learning_rate": 1.9001554864504664e-05, + "loss": 0.1324, "step": 910 }, { - "epoch": 0.05, - "learning_rate": 1.8996057526792163e-05, - "loss": 0.1552, + "epoch": 0.1, + "learning_rate": 1.8996001776988007e-05, + "loss": 0.1101, "step": 915 }, { - "epoch": 0.05, - "learning_rate": 1.899050474762619e-05, - "loss": 0.1581, + "epoch": 0.1, + "learning_rate": 1.899044868947135e-05, + "loss": 0.1102, "step": 920 }, { - "epoch": 0.05, - "learning_rate": 1.8984951968460217e-05, - "loss": 0.1807, + "epoch": 0.1, + "learning_rate": 1.898489560195469e-05, + "loss": 0.0965, "step": 925 }, { - "epoch": 0.05, - "learning_rate": 1.8979399189294243e-05, - "loss": 0.1013, + "epoch": 0.1, + "learning_rate": 1.8979342514438028e-05, + "loss": 0.0451, "step": 930 }, { - "epoch": 0.05, - "learning_rate": 1.897384641012827e-05, - "loss": 0.1408, + "epoch": 0.1, + "learning_rate": 1.897378942692137e-05, + "loss": 0.0892, "step": 935 }, { - "epoch": 0.05, - "learning_rate": 1.8968293630962296e-05, - "loss": 0.1907, + "epoch": 0.1, + "learning_rate": 1.896823633940471e-05, + "loss": 0.1309, "step": 940 }, { - "epoch": 0.05, - "learning_rate": 1.8962740851796325e-05, - "loss": 0.2594, + "epoch": 0.1, + "learning_rate": 1.8962683251888053e-05, + "loss": 0.2524, "step": 945 }, { - "epoch": 0.05, - "learning_rate": 1.8957188072630354e-05, - "loss": 0.4082, + "epoch": 0.11, + "learning_rate": 1.8957130164371392e-05, + "loss": 0.3986, "step": 950 }, { - "epoch": 0.05, - "learning_rate": 1.8951635293464382e-05, - "loss": 0.1815, + "epoch": 0.11, + "learning_rate": 1.8951577076854732e-05, + "loss": 0.0876, "step": 955 }, { - "epoch": 0.05, - "learning_rate": 1.8946082514298408e-05, - "loss": 0.1459, + "epoch": 0.11, + "learning_rate": 1.8946023989338075e-05, + "loss": 0.1454, "step": 960 }, { - "epoch": 0.05, - "learning_rate": 1.8940529735132436e-05, - "loss": 0.1807, + "epoch": 0.11, + "learning_rate": 1.8940470901821414e-05, + "loss": 0.0976, "step": 965 }, { - "epoch": 0.05, - "learning_rate": 1.893497695596646e-05, - "loss": 0.1944, + "epoch": 0.11, + "learning_rate": 1.8934917814304753e-05, + "loss": 0.1096, "step": 970 }, { - "epoch": 0.05, - "learning_rate": 1.892942417680049e-05, - "loss": 0.1371, + "epoch": 0.11, + "learning_rate": 1.8929364726788096e-05, + "loss": 0.0861, "step": 975 }, { - "epoch": 0.05, - "learning_rate": 1.8923871397634515e-05, - "loss": 0.1032, + "epoch": 0.11, + "learning_rate": 1.8923811639271435e-05, + "loss": 0.0918, "step": 980 }, { - "epoch": 0.05, - "learning_rate": 1.8918318618468544e-05, - "loss": 0.104, + "epoch": 0.11, + "learning_rate": 1.8918258551754778e-05, + "loss": 0.0726, "step": 985 }, { - "epoch": 0.05, - "learning_rate": 1.8912765839302573e-05, - "loss": 0.2748, + "epoch": 0.11, + "learning_rate": 1.8912705464238118e-05, + "loss": 0.1302, "step": 990 }, { - "epoch": 0.06, - "learning_rate": 1.89072130601366e-05, - "loss": 0.1682, + "epoch": 0.11, + "learning_rate": 1.8907152376721457e-05, + "loss": 0.1552, "step": 995 }, { - "epoch": 0.06, - "learning_rate": 1.890166028097063e-05, - "loss": 0.4579, + "epoch": 0.11, + "learning_rate": 1.89015992892048e-05, + "loss": 0.2767, "step": 1000 }, { - "epoch": 0.06, - "learning_rate": 1.8896107501804655e-05, - "loss": 0.1335, + "epoch": 0.11, + "learning_rate": 1.889604620168814e-05, + "loss": 0.0916, "step": 1005 }, { - "epoch": 0.06, - "learning_rate": 1.8890554722638684e-05, - "loss": 0.1549, + "epoch": 0.11, + "learning_rate": 1.889049311417148e-05, + "loss": 0.1051, "step": 1010 }, { - "epoch": 0.06, - "learning_rate": 1.888500194347271e-05, - "loss": 0.1343, + "epoch": 0.11, + "learning_rate": 1.888494002665482e-05, + "loss": 0.0975, "step": 1015 }, { - "epoch": 0.06, - "learning_rate": 1.8879449164306738e-05, - "loss": 0.1224, + "epoch": 0.11, + "learning_rate": 1.8879386939138164e-05, + "loss": 0.1404, "step": 1020 }, { - "epoch": 0.06, - "learning_rate": 1.8873896385140763e-05, - "loss": 0.175, + "epoch": 0.11, + "learning_rate": 1.8873833851621503e-05, + "loss": 0.0996, "step": 1025 }, { - "epoch": 0.06, - "learning_rate": 1.8868343605974792e-05, - "loss": 0.1498, + "epoch": 0.11, + "learning_rate": 1.8868280764104846e-05, + "loss": 0.0673, "step": 1030 }, { - "epoch": 0.06, - "learning_rate": 1.886279082680882e-05, - "loss": 0.1116, + "epoch": 0.11, + "learning_rate": 1.8862727676588185e-05, + "loss": 0.0836, "step": 1035 }, { - "epoch": 0.06, - "learning_rate": 1.885723804764285e-05, - "loss": 0.2213, + "epoch": 0.12, + "learning_rate": 1.8857174589071525e-05, + "loss": 0.1149, "step": 1040 }, { - "epoch": 0.06, - "learning_rate": 1.8851685268476874e-05, - "loss": 0.313, + "epoch": 0.12, + "learning_rate": 1.8851621501554868e-05, + "loss": 0.2039, "step": 1045 }, { - "epoch": 0.06, - "learning_rate": 1.8846132489310903e-05, - "loss": 0.4422, + "epoch": 0.12, + "learning_rate": 1.8846068414038207e-05, + "loss": 0.3666, "step": 1050 }, { - "epoch": 0.06, - "learning_rate": 1.8840579710144928e-05, - "loss": 0.1694, + "epoch": 0.12, + "learning_rate": 1.8840515326521546e-05, + "loss": 0.1245, "step": 1055 }, { - "epoch": 0.06, - "learning_rate": 1.8835026930978957e-05, - "loss": 0.1321, + "epoch": 0.12, + "learning_rate": 1.883496223900489e-05, + "loss": 0.1168, "step": 1060 }, { - "epoch": 0.06, - "learning_rate": 1.8829474151812982e-05, - "loss": 0.1274, + "epoch": 0.12, + "learning_rate": 1.882940915148823e-05, + "loss": 0.1029, "step": 1065 }, { - "epoch": 0.06, - "learning_rate": 1.882392137264701e-05, - "loss": 0.1471, + "epoch": 0.12, + "learning_rate": 1.882385606397157e-05, + "loss": 0.1302, "step": 1070 }, { - "epoch": 0.06, - "learning_rate": 1.881836859348104e-05, - "loss": 0.1646, + "epoch": 0.12, + "learning_rate": 1.881830297645491e-05, + "loss": 0.0962, "step": 1075 }, { - "epoch": 0.06, - "learning_rate": 1.8812815814315068e-05, - "loss": 0.0902, + "epoch": 0.12, + "learning_rate": 1.881274988893825e-05, + "loss": 0.0983, "step": 1080 }, { - "epoch": 0.06, - "learning_rate": 1.8807263035149093e-05, - "loss": 0.0842, + "epoch": 0.12, + "learning_rate": 1.8807196801421593e-05, + "loss": 0.1008, "step": 1085 }, { - "epoch": 0.06, - "learning_rate": 1.8801710255983122e-05, - "loss": 0.1957, + "epoch": 0.12, + "learning_rate": 1.8801643713904932e-05, + "loss": 0.1457, "step": 1090 }, { - "epoch": 0.06, - "learning_rate": 1.879615747681715e-05, - "loss": 0.2287, + "epoch": 0.12, + "learning_rate": 1.879609062638827e-05, + "loss": 0.1688, "step": 1095 }, { - "epoch": 0.06, - "learning_rate": 1.8790604697651176e-05, - "loss": 0.3562, + "epoch": 0.12, + "learning_rate": 1.8790537538871614e-05, + "loss": 0.4036, "step": 1100 }, { - "epoch": 0.06, - "learning_rate": 1.8785051918485204e-05, - "loss": 0.1201, + "epoch": 0.12, + "learning_rate": 1.8784984451354954e-05, + "loss": 0.0819, "step": 1105 }, { - "epoch": 0.06, - "learning_rate": 1.877949913931923e-05, - "loss": 0.1578, + "epoch": 0.12, + "learning_rate": 1.8779431363838296e-05, + "loss": 0.113, "step": 1110 }, { - "epoch": 0.06, - "learning_rate": 1.8773946360153258e-05, - "loss": 0.1191, + "epoch": 0.12, + "learning_rate": 1.8773878276321636e-05, + "loss": 0.1456, "step": 1115 }, { - "epoch": 0.06, - "learning_rate": 1.8768393580987287e-05, - "loss": 0.1057, + "epoch": 0.12, + "learning_rate": 1.8768325188804975e-05, + "loss": 0.1137, "step": 1120 }, { - "epoch": 0.06, - "learning_rate": 1.8762840801821312e-05, - "loss": 0.17, + "epoch": 0.12, + "learning_rate": 1.8762772101288318e-05, + "loss": 0.1032, "step": 1125 }, { - "epoch": 0.06, - "learning_rate": 1.875728802265534e-05, - "loss": 0.1386, + "epoch": 0.13, + "learning_rate": 1.875721901377166e-05, + "loss": 0.0595, "step": 1130 }, { - "epoch": 0.06, - "learning_rate": 1.875173524348937e-05, - "loss": 0.0906, + "epoch": 0.13, + "learning_rate": 1.8751665926255e-05, + "loss": 0.0963, "step": 1135 }, { - "epoch": 0.06, - "learning_rate": 1.8746182464323395e-05, - "loss": 0.2144, + "epoch": 0.13, + "learning_rate": 1.874611283873834e-05, + "loss": 0.093, "step": 1140 }, { - "epoch": 0.06, - "learning_rate": 1.8740629685157423e-05, - "loss": 0.2624, + "epoch": 0.13, + "learning_rate": 1.8740559751221682e-05, + "loss": 0.1282, "step": 1145 }, { - "epoch": 0.06, - "learning_rate": 1.873507690599145e-05, - "loss": 0.5116, + "epoch": 0.13, + "learning_rate": 1.873500666370502e-05, + "loss": 0.3287, "step": 1150 }, { - "epoch": 0.06, - "learning_rate": 1.8729524126825477e-05, - "loss": 0.1507, + "epoch": 0.13, + "learning_rate": 1.8729453576188364e-05, + "loss": 0.109, "step": 1155 }, { - "epoch": 0.06, - "learning_rate": 1.8723971347659503e-05, - "loss": 0.124, + "epoch": 0.13, + "learning_rate": 1.8723900488671704e-05, + "loss": 0.093, "step": 1160 }, { - "epoch": 0.06, - "learning_rate": 1.871841856849353e-05, - "loss": 0.1296, + "epoch": 0.13, + "learning_rate": 1.8718347401155043e-05, + "loss": 0.1131, "step": 1165 }, { - "epoch": 0.06, - "learning_rate": 1.871286578932756e-05, - "loss": 0.1664, + "epoch": 0.13, + "learning_rate": 1.8712794313638386e-05, + "loss": 0.1085, "step": 1170 }, { - "epoch": 0.07, - "learning_rate": 1.870731301016159e-05, - "loss": 0.1686, + "epoch": 0.13, + "learning_rate": 1.8707241226121725e-05, + "loss": 0.1316, "step": 1175 }, { - "epoch": 0.07, - "learning_rate": 1.8701760230995617e-05, - "loss": 0.0817, + "epoch": 0.13, + "learning_rate": 1.8701688138605064e-05, + "loss": 0.129, "step": 1180 }, { - "epoch": 0.07, - "learning_rate": 1.8696207451829642e-05, - "loss": 0.1354, + "epoch": 0.13, + "learning_rate": 1.8696135051088407e-05, + "loss": 0.0762, "step": 1185 }, { - "epoch": 0.07, - "learning_rate": 1.869065467266367e-05, - "loss": 0.1889, + "epoch": 0.13, + "learning_rate": 1.8690581963571747e-05, + "loss": 0.1121, "step": 1190 }, { - "epoch": 0.07, - "learning_rate": 1.8685101893497696e-05, - "loss": 0.3455, + "epoch": 0.13, + "learning_rate": 1.868502887605509e-05, + "loss": 0.1596, "step": 1195 }, { - "epoch": 0.07, - "learning_rate": 1.8679549114331725e-05, - "loss": 0.6359, + "epoch": 0.13, + "learning_rate": 1.867947578853843e-05, + "loss": 0.3276, "step": 1200 }, { - "epoch": 0.07, - "learning_rate": 1.867399633516575e-05, - "loss": 0.1781, + "epoch": 0.13, + "learning_rate": 1.8673922701021768e-05, + "loss": 0.1237, "step": 1205 }, { - "epoch": 0.07, - "learning_rate": 1.866844355599978e-05, - "loss": 0.1507, + "epoch": 0.13, + "learning_rate": 1.866836961350511e-05, + "loss": 0.0868, "step": 1210 }, { - "epoch": 0.07, - "learning_rate": 1.8662890776833807e-05, - "loss": 0.1839, + "epoch": 0.13, + "learning_rate": 1.8662816525988454e-05, + "loss": 0.1102, "step": 1215 }, { - "epoch": 0.07, - "learning_rate": 1.8657337997667836e-05, - "loss": 0.1568, + "epoch": 0.14, + "learning_rate": 1.865726343847179e-05, + "loss": 0.1202, "step": 1220 }, { - "epoch": 0.07, - "learning_rate": 1.865178521850186e-05, - "loss": 0.1585, + "epoch": 0.14, + "learning_rate": 1.8651710350955132e-05, + "loss": 0.0971, "step": 1225 }, { - "epoch": 0.07, - "learning_rate": 1.864623243933589e-05, - "loss": 0.1085, + "epoch": 0.14, + "learning_rate": 1.8646157263438475e-05, + "loss": 0.0634, "step": 1230 }, { - "epoch": 0.07, - "learning_rate": 1.8640679660169915e-05, - "loss": 0.1886, + "epoch": 0.14, + "learning_rate": 1.8640604175921814e-05, + "loss": 0.1133, "step": 1235 }, { - "epoch": 0.07, - "learning_rate": 1.8635126881003944e-05, - "loss": 0.1795, + "epoch": 0.14, + "learning_rate": 1.8635051088405154e-05, + "loss": 0.1406, "step": 1240 }, { - "epoch": 0.07, - "learning_rate": 1.862957410183797e-05, - "loss": 0.2619, + "epoch": 0.14, + "learning_rate": 1.8629498000888497e-05, + "loss": 0.1389, "step": 1245 }, { - "epoch": 0.07, - "learning_rate": 1.8624021322671998e-05, - "loss": 0.3515, + "epoch": 0.14, + "learning_rate": 1.8623944913371836e-05, + "loss": 0.3533, "step": 1250 }, { - "epoch": 0.07, - "learning_rate": 1.8618468543506026e-05, - "loss": 0.1733, + "epoch": 0.14, + "learning_rate": 1.861839182585518e-05, + "loss": 0.1312, "step": 1255 }, { - "epoch": 0.07, - "learning_rate": 1.8612915764340055e-05, - "loss": 0.1577, + "epoch": 0.14, + "learning_rate": 1.8612838738338518e-05, + "loss": 0.0902, "step": 1260 }, { - "epoch": 0.07, - "learning_rate": 1.8607362985174084e-05, - "loss": 0.1523, + "epoch": 0.14, + "learning_rate": 1.8607285650821857e-05, + "loss": 0.1013, "step": 1265 }, { - "epoch": 0.07, - "learning_rate": 1.860181020600811e-05, - "loss": 0.2103, + "epoch": 0.14, + "learning_rate": 1.86017325633052e-05, + "loss": 0.1126, "step": 1270 }, { - "epoch": 0.07, - "learning_rate": 1.8596257426842138e-05, - "loss": 0.0489, + "epoch": 0.14, + "learning_rate": 1.859617947578854e-05, + "loss": 0.1037, "step": 1275 }, { - "epoch": 0.07, - "learning_rate": 1.8590704647676163e-05, - "loss": 0.0777, + "epoch": 0.14, + "learning_rate": 1.859062638827188e-05, + "loss": 0.0653, "step": 1280 }, { - "epoch": 0.07, - "learning_rate": 1.858515186851019e-05, - "loss": 0.1812, + "epoch": 0.14, + "learning_rate": 1.858507330075522e-05, + "loss": 0.0894, "step": 1285 }, { - "epoch": 0.07, - "learning_rate": 1.8579599089344217e-05, - "loss": 0.1282, + "epoch": 0.14, + "learning_rate": 1.857952021323856e-05, + "loss": 0.0891, "step": 1290 }, { - "epoch": 0.07, - "learning_rate": 1.8574046310178245e-05, - "loss": 0.2305, + "epoch": 0.14, + "learning_rate": 1.8573967125721904e-05, + "loss": 0.1735, "step": 1295 }, { - "epoch": 0.07, - "learning_rate": 1.8568493531012274e-05, - "loss": 0.4945, + "epoch": 0.14, + "learning_rate": 1.8568414038205243e-05, + "loss": 0.3859, "step": 1300 }, { - "epoch": 0.07, - "learning_rate": 1.8562940751846303e-05, - "loss": 0.1291, + "epoch": 0.14, + "learning_rate": 1.8562860950688583e-05, + "loss": 0.0907, "step": 1305 }, { - "epoch": 0.07, - "learning_rate": 1.8557387972680328e-05, - "loss": 0.1341, + "epoch": 0.15, + "learning_rate": 1.8557307863171925e-05, + "loss": 0.0897, "step": 1310 }, { - "epoch": 0.07, - "learning_rate": 1.8551835193514357e-05, - "loss": 0.1748, + "epoch": 0.15, + "learning_rate": 1.8551754775655265e-05, + "loss": 0.0812, "step": 1315 }, { - "epoch": 0.07, - "learning_rate": 1.8546282414348382e-05, - "loss": 0.1668, + "epoch": 0.15, + "learning_rate": 1.8546201688138607e-05, + "loss": 0.1427, "step": 1320 }, { - "epoch": 0.07, - "learning_rate": 1.854072963518241e-05, - "loss": 0.1381, + "epoch": 0.15, + "learning_rate": 1.8540648600621947e-05, + "loss": 0.0853, "step": 1325 }, { - "epoch": 0.07, - "learning_rate": 1.8535176856016436e-05, - "loss": 0.1074, + "epoch": 0.15, + "learning_rate": 1.853509551310529e-05, + "loss": 0.0821, "step": 1330 }, { - "epoch": 0.07, - "learning_rate": 1.8529624076850464e-05, - "loss": 0.0845, + "epoch": 0.15, + "learning_rate": 1.852954242558863e-05, + "loss": 0.1077, "step": 1335 }, { - "epoch": 0.07, - "learning_rate": 1.8524071297684493e-05, - "loss": 0.1755, + "epoch": 0.15, + "learning_rate": 1.852398933807197e-05, + "loss": 0.1554, "step": 1340 }, { - "epoch": 0.07, - "learning_rate": 1.851851851851852e-05, - "loss": 0.2497, + "epoch": 0.15, + "learning_rate": 1.851843625055531e-05, + "loss": 0.1472, "step": 1345 }, { - "epoch": 0.07, - "learning_rate": 1.8512965739352547e-05, - "loss": 0.3942, + "epoch": 0.15, + "learning_rate": 1.851288316303865e-05, + "loss": 0.3838, "step": 1350 }, { - "epoch": 0.08, - "learning_rate": 1.8507412960186576e-05, - "loss": 0.1222, + "epoch": 0.15, + "learning_rate": 1.8507330075521993e-05, + "loss": 0.1012, "step": 1355 }, { - "epoch": 0.08, - "learning_rate": 1.8501860181020604e-05, - "loss": 0.1453, + "epoch": 0.15, + "learning_rate": 1.8501776988005332e-05, + "loss": 0.0975, "step": 1360 }, { - "epoch": 0.08, - "learning_rate": 1.849630740185463e-05, - "loss": 0.1308, + "epoch": 0.15, + "learning_rate": 1.8496223900488672e-05, + "loss": 0.1068, "step": 1365 }, { - "epoch": 0.08, - "learning_rate": 1.8490754622688658e-05, - "loss": 0.1959, + "epoch": 0.15, + "learning_rate": 1.8490670812972015e-05, + "loss": 0.0981, "step": 1370 }, { - "epoch": 0.08, - "learning_rate": 1.8485201843522683e-05, - "loss": 0.14, + "epoch": 0.15, + "learning_rate": 1.8485117725455354e-05, + "loss": 0.109, "step": 1375 }, { - "epoch": 0.08, - "learning_rate": 1.8479649064356712e-05, - "loss": 0.1304, + "epoch": 0.15, + "learning_rate": 1.8479564637938697e-05, + "loss": 0.1133, "step": 1380 }, { - "epoch": 0.08, - "learning_rate": 1.847409628519074e-05, - "loss": 0.1564, + "epoch": 0.15, + "learning_rate": 1.8474011550422036e-05, + "loss": 0.0995, "step": 1385 }, { - "epoch": 0.08, - "learning_rate": 1.8468543506024766e-05, - "loss": 0.2461, + "epoch": 0.15, + "learning_rate": 1.8468458462905375e-05, + "loss": 0.1472, "step": 1390 }, { - "epoch": 0.08, - "learning_rate": 1.8462990726858795e-05, - "loss": 0.2792, + "epoch": 0.15, + "learning_rate": 1.8462905375388718e-05, + "loss": 0.1471, "step": 1395 }, { - "epoch": 0.08, - "learning_rate": 1.8457437947692823e-05, - "loss": 0.4116, + "epoch": 0.16, + "learning_rate": 1.8457352287872058e-05, + "loss": 0.3345, "step": 1400 }, { - "epoch": 0.08, - "learning_rate": 1.845188516852685e-05, - "loss": 0.1185, + "epoch": 0.16, + "learning_rate": 1.8451799200355397e-05, + "loss": 0.0969, "step": 1405 }, { - "epoch": 0.08, - "learning_rate": 1.8446332389360877e-05, - "loss": 0.1567, + "epoch": 0.16, + "learning_rate": 1.844624611283874e-05, + "loss": 0.0834, "step": 1410 }, { - "epoch": 0.08, - "learning_rate": 1.8440779610194902e-05, - "loss": 0.1063, + "epoch": 0.16, + "learning_rate": 1.844069302532208e-05, + "loss": 0.0904, "step": 1415 }, { - "epoch": 0.08, - "learning_rate": 1.843522683102893e-05, - "loss": 0.184, + "epoch": 0.16, + "learning_rate": 1.8435139937805422e-05, + "loss": 0.1053, "step": 1420 }, { - "epoch": 0.08, - "learning_rate": 1.842967405186296e-05, - "loss": 0.103, + "epoch": 0.16, + "learning_rate": 1.8429586850288765e-05, + "loss": 0.128, "step": 1425 }, { - "epoch": 0.08, - "learning_rate": 1.8424121272696985e-05, - "loss": 0.0841, + "epoch": 0.16, + "learning_rate": 1.84240337627721e-05, + "loss": 0.0652, "step": 1430 }, { - "epoch": 0.08, - "learning_rate": 1.8418568493531014e-05, - "loss": 0.1108, + "epoch": 0.16, + "learning_rate": 1.8418480675255443e-05, + "loss": 0.0541, "step": 1435 }, { - "epoch": 0.08, - "learning_rate": 1.8413015714365042e-05, - "loss": 0.1595, + "epoch": 0.16, + "learning_rate": 1.8412927587738786e-05, + "loss": 0.1123, "step": 1440 }, { - "epoch": 0.08, - "learning_rate": 1.840746293519907e-05, - "loss": 0.2145, + "epoch": 0.16, + "learning_rate": 1.8407374500222125e-05, + "loss": 0.1649, "step": 1445 }, { - "epoch": 0.08, - "learning_rate": 1.8401910156033096e-05, - "loss": 0.369, + "epoch": 0.16, + "learning_rate": 1.8401821412705465e-05, + "loss": 0.3193, "step": 1450 }, { - "epoch": 0.08, - "learning_rate": 1.8396357376867125e-05, - "loss": 0.1741, + "epoch": 0.16, + "learning_rate": 1.8396268325188808e-05, + "loss": 0.1042, "step": 1455 }, { - "epoch": 0.08, - "learning_rate": 1.839080459770115e-05, - "loss": 0.1175, + "epoch": 0.16, + "learning_rate": 1.8390715237672147e-05, + "loss": 0.105, "step": 1460 }, { - "epoch": 0.08, - "learning_rate": 1.838525181853518e-05, - "loss": 0.1004, + "epoch": 0.16, + "learning_rate": 1.838516215015549e-05, + "loss": 0.1443, "step": 1465 }, { - "epoch": 0.08, - "learning_rate": 1.8379699039369204e-05, - "loss": 0.1296, + "epoch": 0.16, + "learning_rate": 1.837960906263883e-05, + "loss": 0.128, "step": 1470 }, { - "epoch": 0.08, - "learning_rate": 1.8374146260203232e-05, - "loss": 0.1752, + "epoch": 0.16, + "learning_rate": 1.837405597512217e-05, + "loss": 0.0748, "step": 1475 }, { - "epoch": 0.08, - "learning_rate": 1.836859348103726e-05, - "loss": 0.0983, + "epoch": 0.16, + "learning_rate": 1.836850288760551e-05, + "loss": 0.0674, "step": 1480 }, { - "epoch": 0.08, - "learning_rate": 1.836304070187129e-05, - "loss": 0.1459, + "epoch": 0.16, + "learning_rate": 1.836294980008885e-05, + "loss": 0.1105, "step": 1485 }, { - "epoch": 0.08, - "learning_rate": 1.8357487922705315e-05, - "loss": 0.2041, + "epoch": 0.17, + "learning_rate": 1.835739671257219e-05, + "loss": 0.1129, "step": 1490 }, { - "epoch": 0.08, - "learning_rate": 1.8351935143539344e-05, - "loss": 0.2331, + "epoch": 0.17, + "learning_rate": 1.8351843625055533e-05, + "loss": 0.183, "step": 1495 }, { - "epoch": 0.08, - "learning_rate": 1.834638236437337e-05, - "loss": 0.6222, + "epoch": 0.17, + "learning_rate": 1.8346290537538872e-05, + "loss": 0.4749, "step": 1500 }, { - "epoch": 0.08, - "learning_rate": 1.8340829585207398e-05, - "loss": 0.1518, + "epoch": 0.17, + "learning_rate": 1.8340737450022215e-05, + "loss": 0.0772, "step": 1505 }, { - "epoch": 0.08, - "learning_rate": 1.8335276806041423e-05, - "loss": 0.136, + "epoch": 0.17, + "learning_rate": 1.8335184362505554e-05, + "loss": 0.1174, "step": 1510 }, { - "epoch": 0.08, - "learning_rate": 1.832972402687545e-05, - "loss": 0.1683, + "epoch": 0.17, + "learning_rate": 1.8329631274988894e-05, + "loss": 0.1242, "step": 1515 }, { - "epoch": 0.08, - "learning_rate": 1.832417124770948e-05, - "loss": 0.1617, + "epoch": 0.17, + "learning_rate": 1.8324078187472236e-05, + "loss": 0.1117, "step": 1520 }, { - "epoch": 0.08, - "learning_rate": 1.831861846854351e-05, - "loss": 0.1483, + "epoch": 0.17, + "learning_rate": 1.831852509995558e-05, + "loss": 0.0696, "step": 1525 }, { - "epoch": 0.08, - "learning_rate": 1.8313065689377537e-05, - "loss": 0.1746, + "epoch": 0.17, + "learning_rate": 1.8312972012438915e-05, + "loss": 0.0599, "step": 1530 }, { - "epoch": 0.09, - "learning_rate": 1.8307512910211563e-05, - "loss": 0.106, + "epoch": 0.17, + "learning_rate": 1.8307418924922258e-05, + "loss": 0.0583, "step": 1535 }, { - "epoch": 0.09, - "learning_rate": 1.830196013104559e-05, - "loss": 0.1278, + "epoch": 0.17, + "learning_rate": 1.83018658374056e-05, + "loss": 0.1187, "step": 1540 }, { - "epoch": 0.09, - "learning_rate": 1.8296407351879617e-05, - "loss": 0.1989, + "epoch": 0.17, + "learning_rate": 1.829631274988894e-05, + "loss": 0.197, "step": 1545 }, { - "epoch": 0.09, - "learning_rate": 1.8290854572713645e-05, - "loss": 0.504, + "epoch": 0.17, + "learning_rate": 1.829075966237228e-05, + "loss": 0.291, "step": 1550 }, { - "epoch": 0.09, - "learning_rate": 1.828530179354767e-05, - "loss": 0.1906, + "epoch": 0.17, + "learning_rate": 1.8285206574855622e-05, + "loss": 0.0863, "step": 1555 }, { - "epoch": 0.09, - "learning_rate": 1.82797490143817e-05, - "loss": 0.1513, + "epoch": 0.17, + "learning_rate": 1.827965348733896e-05, + "loss": 0.1142, "step": 1560 }, { - "epoch": 0.09, - "learning_rate": 1.8274196235215728e-05, - "loss": 0.1459, + "epoch": 0.17, + "learning_rate": 1.8274100399822304e-05, + "loss": 0.099, "step": 1565 }, { - "epoch": 0.09, - "learning_rate": 1.8268643456049756e-05, - "loss": 0.2005, + "epoch": 0.17, + "learning_rate": 1.8268547312305644e-05, + "loss": 0.1095, "step": 1570 }, { - "epoch": 0.09, - "learning_rate": 1.826309067688378e-05, - "loss": 0.1262, + "epoch": 0.17, + "learning_rate": 1.8262994224788983e-05, + "loss": 0.0762, "step": 1575 }, { - "epoch": 0.09, - "learning_rate": 1.825753789771781e-05, - "loss": 0.1131, + "epoch": 0.18, + "learning_rate": 1.8257441137272326e-05, + "loss": 0.0783, "step": 1580 }, { - "epoch": 0.09, - "learning_rate": 1.8251985118551836e-05, - "loss": 0.1688, + "epoch": 0.18, + "learning_rate": 1.8251888049755665e-05, + "loss": 0.0961, "step": 1585 }, { - "epoch": 0.09, - "learning_rate": 1.8246432339385864e-05, - "loss": 0.2079, + "epoch": 0.18, + "learning_rate": 1.8246334962239008e-05, + "loss": 0.1519, "step": 1590 }, { - "epoch": 0.09, - "learning_rate": 1.824087956021989e-05, - "loss": 0.3028, + "epoch": 0.18, + "learning_rate": 1.8240781874722347e-05, + "loss": 0.2982, "step": 1595 }, { - "epoch": 0.09, - "learning_rate": 1.8235326781053918e-05, - "loss": 0.468, + "epoch": 0.18, + "learning_rate": 1.8235228787205687e-05, + "loss": 0.3164, "step": 1600 }, { - "epoch": 0.09, - "learning_rate": 1.8229774001887947e-05, - "loss": 0.1037, - "step": 1605 + "epoch": 0.18, + "learning_rate": 1.822967569968903e-05, + "loss": 0.0642, + "step": 1605 }, { - "epoch": 0.09, - "learning_rate": 1.8224221222721975e-05, - "loss": 0.1726, + "epoch": 0.18, + "learning_rate": 1.822412261217237e-05, + "loss": 0.0919, "step": 1610 }, { - "epoch": 0.09, - "learning_rate": 1.8218668443556e-05, - "loss": 0.1701, + "epoch": 0.18, + "learning_rate": 1.8218569524655708e-05, + "loss": 0.1066, "step": 1615 }, { - "epoch": 0.09, - "learning_rate": 1.821311566439003e-05, - "loss": 0.1554, + "epoch": 0.18, + "learning_rate": 1.821301643713905e-05, + "loss": 0.1179, "step": 1620 }, { - "epoch": 0.09, - "learning_rate": 1.8207562885224058e-05, - "loss": 0.0893, + "epoch": 0.18, + "learning_rate": 1.820746334962239e-05, + "loss": 0.0849, "step": 1625 }, { - "epoch": 0.09, - "learning_rate": 1.8202010106058083e-05, - "loss": 0.1844, + "epoch": 0.18, + "learning_rate": 1.8201910262105733e-05, + "loss": 0.0702, "step": 1630 }, { - "epoch": 0.09, - "learning_rate": 1.8196457326892112e-05, - "loss": 0.1692, + "epoch": 0.18, + "learning_rate": 1.8196357174589072e-05, + "loss": 0.1024, "step": 1635 }, { - "epoch": 0.09, - "learning_rate": 1.8190904547726137e-05, - "loss": 0.1164, + "epoch": 0.18, + "learning_rate": 1.8190804087072415e-05, + "loss": 0.1288, "step": 1640 }, { - "epoch": 0.09, - "learning_rate": 1.8185351768560166e-05, - "loss": 0.2442, + "epoch": 0.18, + "learning_rate": 1.8185250999555754e-05, + "loss": 0.2155, "step": 1645 }, { - "epoch": 0.09, - "learning_rate": 1.8179798989394194e-05, - "loss": 0.5109, + "epoch": 0.18, + "learning_rate": 1.8179697912039097e-05, + "loss": 0.3855, "step": 1650 }, { - "epoch": 0.09, - "learning_rate": 1.8174246210228223e-05, - "loss": 0.2881, + "epoch": 0.18, + "learning_rate": 1.8174144824522437e-05, + "loss": 0.0816, "step": 1655 }, { - "epoch": 0.09, - "learning_rate": 1.8168693431062248e-05, - "loss": 0.1496, + "epoch": 0.18, + "learning_rate": 1.8168591737005776e-05, + "loss": 0.0981, "step": 1660 }, { - "epoch": 0.09, - "learning_rate": 1.8163140651896277e-05, - "loss": 0.1299, + "epoch": 0.18, + "learning_rate": 1.816303864948912e-05, + "loss": 0.1047, "step": 1665 }, { - "epoch": 0.09, - "learning_rate": 1.8157587872730302e-05, - "loss": 0.1583, + "epoch": 0.19, + "learning_rate": 1.8157485561972458e-05, + "loss": 0.1031, "step": 1670 }, { - "epoch": 0.09, - "learning_rate": 1.815203509356433e-05, - "loss": 0.1786, + "epoch": 0.19, + "learning_rate": 1.8151932474455797e-05, + "loss": 0.081, "step": 1675 }, { - "epoch": 0.09, - "learning_rate": 1.8146482314398356e-05, - "loss": 0.0965, + "epoch": 0.19, + "learning_rate": 1.814637938693914e-05, + "loss": 0.0726, "step": 1680 }, { - "epoch": 0.09, - "learning_rate": 1.8140929535232385e-05, - "loss": 0.1091, + "epoch": 0.19, + "learning_rate": 1.814082629942248e-05, + "loss": 0.097, "step": 1685 }, { - "epoch": 0.09, - "learning_rate": 1.8135376756066413e-05, - "loss": 0.1767, + "epoch": 0.19, + "learning_rate": 1.8135273211905822e-05, + "loss": 0.1084, "step": 1690 }, { - "epoch": 0.09, - "learning_rate": 1.8129823976900442e-05, - "loss": 0.3264, + "epoch": 0.19, + "learning_rate": 1.812972012438916e-05, + "loss": 0.1585, "step": 1695 }, { - "epoch": 0.09, - "learning_rate": 1.8124271197734467e-05, - "loss": 0.3536, + "epoch": 0.19, + "learning_rate": 1.81241670368725e-05, + "loss": 0.3579, "step": 1700 }, { - "epoch": 0.09, - "learning_rate": 1.8118718418568496e-05, - "loss": 0.1414, + "epoch": 0.19, + "learning_rate": 1.8118613949355844e-05, + "loss": 0.0689, "step": 1705 }, { - "epoch": 0.09, - "learning_rate": 1.8113165639402524e-05, - "loss": 0.169, + "epoch": 0.19, + "learning_rate": 1.8113060861839183e-05, + "loss": 0.0951, "step": 1710 }, { - "epoch": 0.1, - "learning_rate": 1.810761286023655e-05, - "loss": 0.1481, + "epoch": 0.19, + "learning_rate": 1.8107507774322526e-05, + "loss": 0.0974, "step": 1715 }, { - "epoch": 0.1, - "learning_rate": 1.810206008107058e-05, - "loss": 0.1352, + "epoch": 0.19, + "learning_rate": 1.8101954686805865e-05, + "loss": 0.1083, "step": 1720 }, { - "epoch": 0.1, - "learning_rate": 1.8096507301904604e-05, - "loss": 0.1254, + "epoch": 0.19, + "learning_rate": 1.8096401599289205e-05, + "loss": 0.0747, "step": 1725 }, { - "epoch": 0.1, - "learning_rate": 1.8090954522738632e-05, - "loss": 0.1134, + "epoch": 0.19, + "learning_rate": 1.8090848511772547e-05, + "loss": 0.074, "step": 1730 }, { - "epoch": 0.1, - "learning_rate": 1.8085401743572658e-05, - "loss": 0.1252, + "epoch": 0.19, + "learning_rate": 1.808529542425589e-05, + "loss": 0.0905, "step": 1735 }, { - "epoch": 0.1, - "learning_rate": 1.8079848964406686e-05, - "loss": 0.182, + "epoch": 0.19, + "learning_rate": 1.8079742336739226e-05, + "loss": 0.1717, "step": 1740 }, { - "epoch": 0.1, - "learning_rate": 1.8074296185240715e-05, - "loss": 0.3599, + "epoch": 0.19, + "learning_rate": 1.807418924922257e-05, + "loss": 0.1754, "step": 1745 }, { - "epoch": 0.1, - "learning_rate": 1.8068743406074743e-05, - "loss": 0.4023, + "epoch": 0.19, + "learning_rate": 1.806863616170591e-05, + "loss": 0.4073, "step": 1750 }, { - "epoch": 0.1, - "learning_rate": 1.806319062690877e-05, - "loss": 0.1892, + "epoch": 0.19, + "learning_rate": 1.806308307418925e-05, + "loss": 0.0598, "step": 1755 }, { - "epoch": 0.1, - "learning_rate": 1.8057637847742797e-05, - "loss": 0.1702, + "epoch": 0.2, + "learning_rate": 1.805752998667259e-05, + "loss": 0.0926, "step": 1760 }, { - "epoch": 0.1, - "learning_rate": 1.8052085068576823e-05, - "loss": 0.1621, + "epoch": 0.2, + "learning_rate": 1.8051976899155933e-05, + "loss": 0.0822, "step": 1765 }, { - "epoch": 0.1, - "learning_rate": 1.804653228941085e-05, - "loss": 0.195, + "epoch": 0.2, + "learning_rate": 1.8046423811639272e-05, + "loss": 0.1055, "step": 1770 }, { - "epoch": 0.1, - "learning_rate": 1.8040979510244877e-05, - "loss": 0.1043, + "epoch": 0.2, + "learning_rate": 1.8040870724122615e-05, + "loss": 0.1027, "step": 1775 }, { - "epoch": 0.1, - "learning_rate": 1.8035426731078905e-05, - "loss": 0.1086, + "epoch": 0.2, + "learning_rate": 1.8035317636605955e-05, + "loss": 0.0651, "step": 1780 }, { - "epoch": 0.1, - "learning_rate": 1.8029873951912934e-05, - "loss": 0.089, + "epoch": 0.2, + "learning_rate": 1.8029764549089294e-05, + "loss": 0.0937, "step": 1785 }, { - "epoch": 0.1, - "learning_rate": 1.8024321172746962e-05, - "loss": 0.1475, + "epoch": 0.2, + "learning_rate": 1.8024211461572637e-05, + "loss": 0.1496, "step": 1790 }, { - "epoch": 0.1, - "learning_rate": 1.801876839358099e-05, - "loss": 0.3844, + "epoch": 0.2, + "learning_rate": 1.8018658374055976e-05, + "loss": 0.1447, "step": 1795 }, { - "epoch": 0.1, - "learning_rate": 1.8013215614415016e-05, - "loss": 0.4829, + "epoch": 0.2, + "learning_rate": 1.8014215904042648e-05, + "loss": 0.5813, "step": 1800 }, { - "epoch": 0.1, - "learning_rate": 1.8007662835249045e-05, - "loss": 0.1204, + "epoch": 0.2, + "learning_rate": 1.800866281652599e-05, + "loss": 0.121, "step": 1805 }, { - "epoch": 0.1, - "learning_rate": 1.800211005608307e-05, - "loss": 0.1384, + "epoch": 0.2, + "learning_rate": 1.8003109729009333e-05, + "loss": 0.1069, "step": 1810 }, { - "epoch": 0.1, - "learning_rate": 1.79965572769171e-05, - "loss": 0.143, + "epoch": 0.2, + "learning_rate": 1.799755664149267e-05, + "loss": 0.0923, "step": 1815 }, { - "epoch": 0.1, - "learning_rate": 1.7991004497751124e-05, - "loss": 0.1422, + "epoch": 0.2, + "learning_rate": 1.7992003553976012e-05, + "loss": 0.12, "step": 1820 }, { - "epoch": 0.1, - "learning_rate": 1.7985451718585153e-05, - "loss": 0.077, + "epoch": 0.2, + "learning_rate": 1.7986450466459355e-05, + "loss": 0.0846, "step": 1825 }, { - "epoch": 0.1, - "learning_rate": 1.797989893941918e-05, - "loss": 0.0444, + "epoch": 0.2, + "learning_rate": 1.7980897378942694e-05, + "loss": 0.1039, "step": 1830 }, { - "epoch": 0.1, - "learning_rate": 1.797434616025321e-05, - "loss": 0.0705, + "epoch": 0.2, + "learning_rate": 1.7975344291426033e-05, + "loss": 0.1386, "step": 1835 }, { - "epoch": 0.1, - "learning_rate": 1.7968793381087235e-05, - "loss": 0.1913, + "epoch": 0.2, + "learning_rate": 1.7969791203909376e-05, + "loss": 0.1327, "step": 1840 }, { - "epoch": 0.1, - "learning_rate": 1.7963240601921264e-05, - "loss": 0.1767, + "epoch": 0.2, + "learning_rate": 1.7964238116392715e-05, + "loss": 0.1693, "step": 1845 }, { - "epoch": 0.1, - "learning_rate": 1.795768782275529e-05, - "loss": 0.5094, + "epoch": 0.21, + "learning_rate": 1.7958685028876058e-05, + "loss": 0.4163, "step": 1850 }, { - "epoch": 0.1, - "learning_rate": 1.7952135043589318e-05, - "loss": 0.1402, + "epoch": 0.21, + "learning_rate": 1.7953131941359398e-05, + "loss": 0.0809, "step": 1855 }, { - "epoch": 0.1, - "learning_rate": 1.7946582264423343e-05, - "loss": 0.2011, + "epoch": 0.21, + "learning_rate": 1.7947578853842737e-05, + "loss": 0.1079, "step": 1860 }, { - "epoch": 0.1, - "learning_rate": 1.7941029485257372e-05, - "loss": 0.3162, + "epoch": 0.21, + "learning_rate": 1.794202576632608e-05, + "loss": 0.133, "step": 1865 }, { - "epoch": 0.1, - "learning_rate": 1.79354767060914e-05, - "loss": 0.1685, + "epoch": 0.21, + "learning_rate": 1.793647267880942e-05, + "loss": 0.1613, "step": 1870 }, { - "epoch": 0.1, - "learning_rate": 1.792992392692543e-05, - "loss": 0.1241, + "epoch": 0.21, + "learning_rate": 1.793091959129276e-05, + "loss": 0.0594, "step": 1875 }, { - "epoch": 0.1, - "learning_rate": 1.7924371147759454e-05, - "loss": 0.0725, + "epoch": 0.21, + "learning_rate": 1.79253665037761e-05, + "loss": 0.0836, "step": 1880 }, { - "epoch": 0.1, - "learning_rate": 1.7918818368593483e-05, - "loss": 0.113, + "epoch": 0.21, + "learning_rate": 1.791981341625944e-05, + "loss": 0.1182, "step": 1885 }, { - "epoch": 0.1, - "learning_rate": 1.791326558942751e-05, - "loss": 0.1232, + "epoch": 0.21, + "learning_rate": 1.7914260328742783e-05, + "loss": 0.0962, "step": 1890 }, { - "epoch": 0.11, - "learning_rate": 1.7907712810261537e-05, - "loss": 0.2065, + "epoch": 0.21, + "learning_rate": 1.7908707241226123e-05, + "loss": 0.1491, "step": 1895 }, { - "epoch": 0.11, - "learning_rate": 1.7902160031095565e-05, - "loss": 0.3366, + "epoch": 0.21, + "learning_rate": 1.7903154153709462e-05, + "loss": 0.4326, "step": 1900 }, { - "epoch": 0.11, - "learning_rate": 1.789660725192959e-05, - "loss": 0.1323, + "epoch": 0.21, + "learning_rate": 1.7897601066192805e-05, + "loss": 0.0958, "step": 1905 }, { - "epoch": 0.11, - "learning_rate": 1.789105447276362e-05, - "loss": 0.1542, + "epoch": 0.21, + "learning_rate": 1.7892047978676144e-05, + "loss": 0.1066, "step": 1910 }, { - "epoch": 0.11, - "learning_rate": 1.7885501693597648e-05, - "loss": 0.1622, + "epoch": 0.21, + "learning_rate": 1.7886494891159487e-05, + "loss": 0.1029, "step": 1915 }, { - "epoch": 0.11, - "learning_rate": 1.7879948914431677e-05, - "loss": 0.1642, + "epoch": 0.21, + "learning_rate": 1.7880941803642826e-05, + "loss": 0.0875, "step": 1920 }, { - "epoch": 0.11, - "learning_rate": 1.7874396135265702e-05, - "loss": 0.0918, + "epoch": 0.21, + "learning_rate": 1.787538871612617e-05, + "loss": 0.0837, "step": 1925 }, { - "epoch": 0.11, - "learning_rate": 1.786884335609973e-05, - "loss": 0.1178, + "epoch": 0.21, + "learning_rate": 1.786983562860951e-05, + "loss": 0.0565, "step": 1930 }, { - "epoch": 0.11, - "learning_rate": 1.7863290576933756e-05, - "loss": 0.1393, + "epoch": 0.21, + "learning_rate": 1.786428254109285e-05, + "loss": 0.1131, "step": 1935 }, { - "epoch": 0.11, - "learning_rate": 1.7857737797767784e-05, - "loss": 0.1588, + "epoch": 0.22, + "learning_rate": 1.785872945357619e-05, + "loss": 0.1341, "step": 1940 }, { - "epoch": 0.11, - "learning_rate": 1.785218501860181e-05, - "loss": 0.2602, + "epoch": 0.22, + "learning_rate": 1.785317636605953e-05, + "loss": 0.1491, "step": 1945 }, { - "epoch": 0.11, - "learning_rate": 1.784663223943584e-05, - "loss": 0.3922, + "epoch": 0.22, + "learning_rate": 1.7847623278542873e-05, + "loss": 0.3896, "step": 1950 }, { - "epoch": 0.11, - "learning_rate": 1.7841079460269867e-05, - "loss": 0.1129, + "epoch": 0.22, + "learning_rate": 1.7842070191026212e-05, + "loss": 0.075, "step": 1955 }, { - "epoch": 0.11, - "learning_rate": 1.7835526681103896e-05, - "loss": 0.1722, + "epoch": 0.22, + "learning_rate": 1.783651710350955e-05, + "loss": 0.1017, "step": 1960 }, { - "epoch": 0.11, - "learning_rate": 1.782997390193792e-05, - "loss": 0.1399, + "epoch": 0.22, + "learning_rate": 1.7830964015992894e-05, + "loss": 0.1047, "step": 1965 }, { - "epoch": 0.11, - "learning_rate": 1.782442112277195e-05, - "loss": 0.1258, + "epoch": 0.22, + "learning_rate": 1.7825410928476234e-05, + "loss": 0.0862, "step": 1970 }, { - "epoch": 0.11, - "learning_rate": 1.7818868343605978e-05, - "loss": 0.1405, + "epoch": 0.22, + "learning_rate": 1.7819857840959576e-05, + "loss": 0.1087, "step": 1975 }, { - "epoch": 0.11, - "learning_rate": 1.7813315564440003e-05, - "loss": 0.0442, + "epoch": 0.22, + "learning_rate": 1.7814304753442916e-05, + "loss": 0.0662, "step": 1980 }, { - "epoch": 0.11, - "learning_rate": 1.7807762785274032e-05, - "loss": 0.1456, + "epoch": 0.22, + "learning_rate": 1.7808751665926255e-05, + "loss": 0.0751, "step": 1985 }, { - "epoch": 0.11, - "learning_rate": 1.7802210006108057e-05, - "loss": 0.12, + "epoch": 0.22, + "learning_rate": 1.7803198578409598e-05, + "loss": 0.1606, "step": 1990 }, { - "epoch": 0.11, - "learning_rate": 1.7796657226942086e-05, - "loss": 0.2641, + "epoch": 0.22, + "learning_rate": 1.7797645490892937e-05, + "loss": 0.184, "step": 1995 }, { - "epoch": 0.11, - "learning_rate": 1.7791104447776115e-05, - "loss": 0.4103, + "epoch": 0.22, + "learning_rate": 1.7792092403376277e-05, + "loss": 0.3407, "step": 2000 }, { - "epoch": 0.11, - "eval_loss": 0.06214262545108795, - "eval_runtime": 1713.2184, - "eval_samples_per_second": 9.35, + "epoch": 0.22, + "eval_loss": 0.050165969878435135, + "eval_runtime": 1712.7733, + "eval_samples_per_second": 9.352, "eval_steps_per_second": 2.338, - "eval_wer": 0.18781103498016588, + "eval_wer": 0.13122971510998918, "step": 2000 }, { - "epoch": 0.11, - "learning_rate": 1.778555166861014e-05, - "loss": 0.1797, + "epoch": 0.22, + "learning_rate": 1.778653931585962e-05, + "loss": 0.1099, "step": 2005 }, { - "epoch": 0.11, - "learning_rate": 1.777999888944417e-05, - "loss": 0.1758, + "epoch": 0.22, + "learning_rate": 1.778098622834296e-05, + "loss": 0.1189, "step": 2010 }, { - "epoch": 0.11, - "learning_rate": 1.7774446110278197e-05, - "loss": 0.1448, + "epoch": 0.22, + "learning_rate": 1.77754331408263e-05, + "loss": 0.0952, "step": 2015 }, { - "epoch": 0.11, - "learning_rate": 1.7768893331112222e-05, - "loss": 0.1724, + "epoch": 0.22, + "learning_rate": 1.7769880053309644e-05, + "loss": 0.0993, "step": 2020 }, { - "epoch": 0.11, - "learning_rate": 1.776334055194625e-05, - "loss": 0.1515, + "epoch": 0.22, + "learning_rate": 1.776432696579298e-05, + "loss": 0.1227, "step": 2025 }, { - "epoch": 0.11, - "learning_rate": 1.7757787772780276e-05, - "loss": 0.055, + "epoch": 0.23, + "learning_rate": 1.7758773878276323e-05, + "loss": 0.0403, "step": 2030 }, { - "epoch": 0.11, - "learning_rate": 1.7752234993614305e-05, - "loss": 0.1555, + "epoch": 0.23, + "learning_rate": 1.7753220790759666e-05, + "loss": 0.0698, "step": 2035 }, { - "epoch": 0.11, - "learning_rate": 1.7746682214448334e-05, - "loss": 0.1705, + "epoch": 0.23, + "learning_rate": 1.7747667703243005e-05, + "loss": 0.1324, "step": 2040 }, { - "epoch": 0.11, - "learning_rate": 1.774112943528236e-05, - "loss": 0.1809, + "epoch": 0.23, + "learning_rate": 1.7742114615726344e-05, + "loss": 0.1785, "step": 2045 }, { - "epoch": 0.11, - "learning_rate": 1.7735576656116388e-05, - "loss": 0.3863, + "epoch": 0.23, + "learning_rate": 1.7736561528209687e-05, + "loss": 0.2467, "step": 2050 }, { - "epoch": 0.11, - "learning_rate": 1.7730023876950416e-05, - "loss": 0.1226, + "epoch": 0.23, + "learning_rate": 1.7731008440693027e-05, + "loss": 0.0919, "step": 2055 }, { - "epoch": 0.11, - "learning_rate": 1.7724471097784445e-05, - "loss": 0.1277, + "epoch": 0.23, + "learning_rate": 1.772545535317637e-05, + "loss": 0.0911, "step": 2060 }, { - "epoch": 0.11, - "learning_rate": 1.771891831861847e-05, - "loss": 0.1516, + "epoch": 0.23, + "learning_rate": 1.771990226565971e-05, + "loss": 0.098, "step": 2065 }, { - "epoch": 0.11, - "learning_rate": 1.77133655394525e-05, - "loss": 0.1531, + "epoch": 0.23, + "learning_rate": 1.7714349178143048e-05, + "loss": 0.1045, "step": 2070 }, { - "epoch": 0.12, - "learning_rate": 1.7707812760286524e-05, - "loss": 0.1011, + "epoch": 0.23, + "learning_rate": 1.770879609062639e-05, + "loss": 0.09, "step": 2075 }, { - "epoch": 0.12, - "learning_rate": 1.7702259981120553e-05, - "loss": 0.1175, + "epoch": 0.23, + "learning_rate": 1.770324300310973e-05, + "loss": 0.0827, "step": 2080 }, { - "epoch": 0.12, - "learning_rate": 1.7696707201954578e-05, - "loss": 0.0744, + "epoch": 0.23, + "learning_rate": 1.769768991559307e-05, + "loss": 0.072, "step": 2085 }, { - "epoch": 0.12, - "learning_rate": 1.7691154422788606e-05, - "loss": 0.1591, + "epoch": 0.23, + "learning_rate": 1.7692136828076412e-05, + "loss": 0.0998, "step": 2090 }, { - "epoch": 0.12, - "learning_rate": 1.7685601643622635e-05, - "loss": 0.2806, + "epoch": 0.23, + "learning_rate": 1.768658374055975e-05, + "loss": 0.205, "step": 2095 }, { - "epoch": 0.12, - "learning_rate": 1.7680048864456664e-05, - "loss": 0.3483, + "epoch": 0.23, + "learning_rate": 1.7681030653043094e-05, + "loss": 0.3571, "step": 2100 }, { - "epoch": 0.12, - "learning_rate": 1.767449608529069e-05, - "loss": 0.1594, + "epoch": 0.23, + "learning_rate": 1.7675477565526434e-05, + "loss": 0.122, "step": 2105 }, { - "epoch": 0.12, - "learning_rate": 1.7668943306124718e-05, - "loss": 0.1872, + "epoch": 0.23, + "learning_rate": 1.7669924478009773e-05, + "loss": 0.1288, "step": 2110 }, { - "epoch": 0.12, - "learning_rate": 1.7663390526958743e-05, - "loss": 0.1682, + "epoch": 0.23, + "learning_rate": 1.7664371390493116e-05, + "loss": 0.1193, "step": 2115 }, { - "epoch": 0.12, - "learning_rate": 1.765783774779277e-05, - "loss": 0.1582, + "epoch": 0.24, + "learning_rate": 1.765881830297646e-05, + "loss": 0.1113, "step": 2120 }, { - "epoch": 0.12, - "learning_rate": 1.7652284968626797e-05, - "loss": 0.1726, + "epoch": 0.24, + "learning_rate": 1.7653265215459795e-05, + "loss": 0.102, "step": 2125 }, { - "epoch": 0.12, - "learning_rate": 1.7646732189460825e-05, - "loss": 0.0793, + "epoch": 0.24, + "learning_rate": 1.7647712127943137e-05, + "loss": 0.038, "step": 2130 }, { - "epoch": 0.12, - "learning_rate": 1.7641179410294854e-05, - "loss": 0.0958, + "epoch": 0.24, + "learning_rate": 1.764215904042648e-05, + "loss": 0.0566, "step": 2135 }, { - "epoch": 0.12, - "learning_rate": 1.7635626631128883e-05, - "loss": 0.2276, + "epoch": 0.24, + "learning_rate": 1.763660595290982e-05, + "loss": 0.0881, "step": 2140 }, { - "epoch": 0.12, - "learning_rate": 1.7630073851962908e-05, - "loss": 0.1913, + "epoch": 0.24, + "learning_rate": 1.763105286539316e-05, + "loss": 0.1674, "step": 2145 }, { - "epoch": 0.12, - "learning_rate": 1.7624521072796937e-05, - "loss": 0.4225, + "epoch": 0.24, + "learning_rate": 1.76254997778765e-05, + "loss": 0.2494, "step": 2150 }, { - "epoch": 0.12, - "learning_rate": 1.7618968293630965e-05, - "loss": 0.1837, + "epoch": 0.24, + "learning_rate": 1.761994669035984e-05, + "loss": 0.0807, "step": 2155 }, { - "epoch": 0.12, - "learning_rate": 1.761341551446499e-05, - "loss": 0.1323, + "epoch": 0.24, + "learning_rate": 1.7614393602843184e-05, + "loss": 0.0811, "step": 2160 }, { - "epoch": 0.12, - "learning_rate": 1.760786273529902e-05, - "loss": 0.1299, + "epoch": 0.24, + "learning_rate": 1.7608840515326523e-05, + "loss": 0.117, "step": 2165 }, { - "epoch": 0.12, - "learning_rate": 1.7602309956133044e-05, - "loss": 0.1072, + "epoch": 0.24, + "learning_rate": 1.7603287427809863e-05, + "loss": 0.1444, "step": 2170 }, { - "epoch": 0.12, - "learning_rate": 1.7596757176967073e-05, - "loss": 0.1054, + "epoch": 0.24, + "learning_rate": 1.7597734340293205e-05, + "loss": 0.0817, "step": 2175 }, { - "epoch": 0.12, - "learning_rate": 1.7591204397801102e-05, - "loss": 0.0669, + "epoch": 0.24, + "learning_rate": 1.7592181252776545e-05, + "loss": 0.0418, "step": 2180 }, { - "epoch": 0.12, - "learning_rate": 1.758565161863513e-05, - "loss": 0.149, + "epoch": 0.24, + "learning_rate": 1.7586628165259887e-05, + "loss": 0.1151, "step": 2185 }, { - "epoch": 0.12, - "learning_rate": 1.7580098839469156e-05, - "loss": 0.2359, + "epoch": 0.24, + "learning_rate": 1.7581075077743227e-05, + "loss": 0.0961, "step": 2190 }, { - "epoch": 0.12, - "learning_rate": 1.7574546060303184e-05, - "loss": 0.2178, + "epoch": 0.24, + "learning_rate": 1.7575521990226566e-05, + "loss": 0.1609, "step": 2195 }, { - "epoch": 0.12, - "learning_rate": 1.756899328113721e-05, - "loss": 0.5178, + "epoch": 0.24, + "learning_rate": 1.756996890270991e-05, + "loss": 0.3116, "step": 2200 }, { - "epoch": 0.12, - "learning_rate": 1.7563440501971238e-05, - "loss": 0.129, + "epoch": 0.24, + "learning_rate": 1.7564415815193248e-05, + "loss": 0.0887, "step": 2205 }, { - "epoch": 0.12, - "learning_rate": 1.7557887722805263e-05, - "loss": 0.1416, + "epoch": 0.25, + "learning_rate": 1.7558862727676588e-05, + "loss": 0.087, "step": 2210 }, { - "epoch": 0.12, - "learning_rate": 1.7552334943639292e-05, - "loss": 0.156, + "epoch": 0.25, + "learning_rate": 1.755330964015993e-05, + "loss": 0.0946, "step": 2215 }, { - "epoch": 0.12, - "learning_rate": 1.754678216447332e-05, - "loss": 0.1461, + "epoch": 0.25, + "learning_rate": 1.754775655264327e-05, + "loss": 0.1125, "step": 2220 }, { - "epoch": 0.12, - "learning_rate": 1.754122938530735e-05, - "loss": 0.1472, + "epoch": 0.25, + "learning_rate": 1.7542203465126612e-05, + "loss": 0.0796, "step": 2225 }, { - "epoch": 0.12, - "learning_rate": 1.7535676606141375e-05, - "loss": 0.0692, + "epoch": 0.25, + "learning_rate": 1.7536650377609952e-05, + "loss": 0.0459, "step": 2230 }, { - "epoch": 0.12, - "learning_rate": 1.7530123826975403e-05, - "loss": 0.1077, + "epoch": 0.25, + "learning_rate": 1.7531097290093295e-05, + "loss": 0.0971, "step": 2235 }, { - "epoch": 0.12, - "learning_rate": 1.7524571047809432e-05, - "loss": 0.1522, + "epoch": 0.25, + "learning_rate": 1.7525544202576634e-05, + "loss": 0.1492, "step": 2240 }, { - "epoch": 0.12, - "learning_rate": 1.7519018268643457e-05, - "loss": 0.2545, + "epoch": 0.25, + "learning_rate": 1.7519991115059977e-05, + "loss": 0.171, "step": 2245 }, { - "epoch": 0.12, - "learning_rate": 1.7513465489477486e-05, - "loss": 0.3949, + "epoch": 0.25, + "learning_rate": 1.7514438027543316e-05, + "loss": 0.3832, "step": 2250 }, { - "epoch": 0.13, - "learning_rate": 1.750791271031151e-05, - "loss": 0.1139, + "epoch": 0.25, + "learning_rate": 1.7508884940026655e-05, + "loss": 0.0961, "step": 2255 }, { - "epoch": 0.13, - "learning_rate": 1.750235993114554e-05, - "loss": 0.1266, + "epoch": 0.25, + "learning_rate": 1.7503331852509998e-05, + "loss": 0.0928, "step": 2260 }, { - "epoch": 0.13, - "learning_rate": 1.749680715197957e-05, - "loss": 0.1431, + "epoch": 0.25, + "learning_rate": 1.7497778764993338e-05, + "loss": 0.1037, "step": 2265 }, { - "epoch": 0.13, - "learning_rate": 1.7491254372813597e-05, - "loss": 0.1993, + "epoch": 0.25, + "learning_rate": 1.7492225677476677e-05, + "loss": 0.1006, "step": 2270 }, { - "epoch": 0.13, - "learning_rate": 1.7485701593647622e-05, - "loss": 0.1558, + "epoch": 0.25, + "learning_rate": 1.748667258996002e-05, + "loss": 0.1022, "step": 2275 }, { - "epoch": 0.13, - "learning_rate": 1.748014881448165e-05, - "loss": 0.1741, + "epoch": 0.25, + "learning_rate": 1.748111950244336e-05, + "loss": 0.0412, "step": 2280 }, { - "epoch": 0.13, - "learning_rate": 1.7474596035315676e-05, - "loss": 0.1423, + "epoch": 0.25, + "learning_rate": 1.7475566414926702e-05, + "loss": 0.1383, "step": 2285 }, { - "epoch": 0.13, - "learning_rate": 1.7469043256149705e-05, - "loss": 0.1417, + "epoch": 0.25, + "learning_rate": 1.747001332741004e-05, + "loss": 0.1163, "step": 2290 }, { - "epoch": 0.13, - "learning_rate": 1.746349047698373e-05, - "loss": 0.1437, + "epoch": 0.25, + "learning_rate": 1.746446023989338e-05, + "loss": 0.1866, "step": 2295 }, { - "epoch": 0.13, - "learning_rate": 1.745793769781776e-05, - "loss": 0.4293, + "epoch": 0.26, + "learning_rate": 1.7458907152376723e-05, + "loss": 0.2643, "step": 2300 }, { - "epoch": 0.13, - "learning_rate": 1.7452384918651787e-05, - "loss": 0.1076, + "epoch": 0.26, + "learning_rate": 1.7453354064860063e-05, + "loss": 0.0854, "step": 2305 }, { - "epoch": 0.13, - "learning_rate": 1.7446832139485813e-05, - "loss": 0.1443, + "epoch": 0.26, + "learning_rate": 1.7447800977343405e-05, + "loss": 0.0796, "step": 2310 }, { - "epoch": 0.13, - "learning_rate": 1.744127936031984e-05, - "loss": 0.1416, + "epoch": 0.26, + "learning_rate": 1.7442247889826745e-05, + "loss": 0.1129, "step": 2315 }, { - "epoch": 0.13, - "learning_rate": 1.743572658115387e-05, - "loss": 0.1389, + "epoch": 0.26, + "learning_rate": 1.7436694802310084e-05, + "loss": 0.0971, "step": 2320 }, { - "epoch": 0.13, - "learning_rate": 1.74301738019879e-05, - "loss": 0.2184, + "epoch": 0.26, + "learning_rate": 1.7431141714793427e-05, + "loss": 0.0693, "step": 2325 }, { - "epoch": 0.13, - "learning_rate": 1.7424621022821924e-05, - "loss": 0.1267, + "epoch": 0.26, + "learning_rate": 1.742558862727677e-05, + "loss": 0.0716, "step": 2330 }, { - "epoch": 0.13, - "learning_rate": 1.7419068243655952e-05, - "loss": 0.1047, + "epoch": 0.26, + "learning_rate": 1.7420035539760106e-05, + "loss": 0.1012, "step": 2335 }, { - "epoch": 0.13, - "learning_rate": 1.7413515464489978e-05, - "loss": 0.1697, + "epoch": 0.26, + "learning_rate": 1.741448245224345e-05, + "loss": 0.1319, "step": 2340 }, { - "epoch": 0.13, - "learning_rate": 1.7407962685324006e-05, - "loss": 0.1957, + "epoch": 0.26, + "learning_rate": 1.740892936472679e-05, + "loss": 0.2252, "step": 2345 }, { - "epoch": 0.13, - "learning_rate": 1.740240990615803e-05, - "loss": 0.3344, + "epoch": 0.26, + "learning_rate": 1.740337627721013e-05, + "loss": 0.2975, "step": 2350 }, { - "epoch": 0.13, - "learning_rate": 1.739685712699206e-05, - "loss": 0.1444, + "epoch": 0.26, + "learning_rate": 1.739782318969347e-05, + "loss": 0.0746, "step": 2355 }, { - "epoch": 0.13, - "learning_rate": 1.739130434782609e-05, - "loss": 0.1352, + "epoch": 0.26, + "learning_rate": 1.7392270102176813e-05, + "loss": 0.0867, "step": 2360 }, { - "epoch": 0.13, - "learning_rate": 1.7385751568660117e-05, - "loss": 0.1475, + "epoch": 0.26, + "learning_rate": 1.7386717014660152e-05, + "loss": 0.1057, "step": 2365 }, { - "epoch": 0.13, - "learning_rate": 1.7380198789494143e-05, - "loss": 0.1388, + "epoch": 0.26, + "learning_rate": 1.7381163927143495e-05, + "loss": 0.1079, "step": 2370 }, { - "epoch": 0.13, - "learning_rate": 1.737464601032817e-05, - "loss": 0.1004, + "epoch": 0.26, + "learning_rate": 1.7375610839626834e-05, + "loss": 0.0656, "step": 2375 }, { - "epoch": 0.13, - "learning_rate": 1.7369093231162197e-05, - "loss": 0.0734, - "step": 2380 + "epoch": 0.26, + "learning_rate": 1.7370057752110174e-05, + "loss": 0.077, + "step": 2380 }, { - "epoch": 0.13, - "learning_rate": 1.7363540451996225e-05, - "loss": 0.1189, + "epoch": 0.26, + "learning_rate": 1.7364504664593516e-05, + "loss": 0.1688, "step": 2385 }, { - "epoch": 0.13, - "learning_rate": 1.735798767283025e-05, - "loss": 0.204, + "epoch": 0.27, + "learning_rate": 1.7358951577076856e-05, + "loss": 0.1224, "step": 2390 }, { - "epoch": 0.13, - "learning_rate": 1.735243489366428e-05, - "loss": 0.1835, + "epoch": 0.27, + "learning_rate": 1.7353398489560195e-05, + "loss": 0.1352, "step": 2395 }, { - "epoch": 0.13, - "learning_rate": 1.7346882114498308e-05, - "loss": 0.3966, + "epoch": 0.27, + "learning_rate": 1.7347845402043538e-05, + "loss": 0.2779, "step": 2400 }, { - "epoch": 0.13, - "learning_rate": 1.7341329335332336e-05, - "loss": 0.1322, + "epoch": 0.27, + "learning_rate": 1.7342292314526877e-05, + "loss": 0.0789, "step": 2405 }, { - "epoch": 0.13, - "learning_rate": 1.7335776556166365e-05, - "loss": 0.1389, + "epoch": 0.27, + "learning_rate": 1.733673922701022e-05, + "loss": 0.0798, "step": 2410 }, { - "epoch": 0.13, - "learning_rate": 1.733022377700039e-05, - "loss": 0.1493, + "epoch": 0.27, + "learning_rate": 1.733118613949356e-05, + "loss": 0.088, "step": 2415 }, { - "epoch": 0.13, - "learning_rate": 1.732467099783442e-05, - "loss": 0.1795, + "epoch": 0.27, + "learning_rate": 1.73256330519769e-05, + "loss": 0.0995, "step": 2420 }, { - "epoch": 0.13, - "learning_rate": 1.7319118218668444e-05, - "loss": 0.1456, + "epoch": 0.27, + "learning_rate": 1.732007996446024e-05, + "loss": 0.1002, "step": 2425 }, { - "epoch": 0.13, - "learning_rate": 1.7313565439502473e-05, - "loss": 0.0746, + "epoch": 0.27, + "learning_rate": 1.7314526876943584e-05, + "loss": 0.0484, "step": 2430 }, { - "epoch": 0.14, - "learning_rate": 1.7308012660336498e-05, - "loss": 0.075, + "epoch": 0.27, + "learning_rate": 1.7308973789426924e-05, + "loss": 0.0947, "step": 2435 }, { - "epoch": 0.14, - "learning_rate": 1.7302459881170527e-05, - "loss": 0.1138, + "epoch": 0.27, + "learning_rate": 1.7303420701910263e-05, + "loss": 0.121, "step": 2440 }, { - "epoch": 0.14, - "learning_rate": 1.7296907102004555e-05, - "loss": 0.178, + "epoch": 0.27, + "learning_rate": 1.7297867614393606e-05, + "loss": 0.2656, "step": 2445 }, { - "epoch": 0.14, - "learning_rate": 1.729246487867178e-05, - "loss": 0.4019, + "epoch": 0.27, + "learning_rate": 1.7292314526876945e-05, + "loss": 0.3153, "step": 2450 }, { - "epoch": 0.14, - "learning_rate": 1.7286912099505804e-05, - "loss": 0.2931, + "epoch": 0.27, + "learning_rate": 1.7286761439360288e-05, + "loss": 0.1041, "step": 2455 }, { - "epoch": 0.14, - "learning_rate": 1.7281359320339832e-05, - "loss": 0.1418, + "epoch": 0.27, + "learning_rate": 1.7281208351843627e-05, + "loss": 0.1142, "step": 2460 }, { - "epoch": 0.14, - "learning_rate": 1.7275806541173857e-05, - "loss": 0.1609, + "epoch": 0.27, + "learning_rate": 1.7275655264326967e-05, + "loss": 0.0924, "step": 2465 }, { - "epoch": 0.14, - "learning_rate": 1.7270253762007886e-05, - "loss": 0.1588, + "epoch": 0.27, + "learning_rate": 1.727010217681031e-05, + "loss": 0.1132, "step": 2470 }, { - "epoch": 0.14, - "learning_rate": 1.7264700982841915e-05, - "loss": 0.1255, + "epoch": 0.27, + "learning_rate": 1.726454908929365e-05, + "loss": 0.1095, "step": 2475 }, { - "epoch": 0.14, - "learning_rate": 1.725914820367594e-05, - "loss": 0.1065, + "epoch": 0.28, + "learning_rate": 1.7258996001776988e-05, + "loss": 0.0713, "step": 2480 }, { - "epoch": 0.14, - "learning_rate": 1.725359542450997e-05, - "loss": 0.1305, + "epoch": 0.28, + "learning_rate": 1.725344291426033e-05, + "loss": 0.1302, "step": 2485 }, { - "epoch": 0.14, - "learning_rate": 1.7248042645343997e-05, - "loss": 0.2013, + "epoch": 0.28, + "learning_rate": 1.724788982674367e-05, + "loss": 0.1497, "step": 2490 }, { - "epoch": 0.14, - "learning_rate": 1.7242489866178023e-05, - "loss": 0.2284, + "epoch": 0.28, + "learning_rate": 1.7242336739227013e-05, + "loss": 0.1711, "step": 2495 }, { - "epoch": 0.14, - "learning_rate": 1.723693708701205e-05, - "loss": 0.4172, + "epoch": 0.28, + "learning_rate": 1.7236783651710352e-05, + "loss": 0.3014, "step": 2500 }, { - "epoch": 0.14, - "learning_rate": 1.7231384307846076e-05, - "loss": 0.1652, + "epoch": 0.28, + "learning_rate": 1.723123056419369e-05, + "loss": 0.086, "step": 2505 }, { - "epoch": 0.14, - "learning_rate": 1.7225831528680105e-05, - "loss": 0.1251, + "epoch": 0.28, + "learning_rate": 1.7225677476677034e-05, + "loss": 0.0806, "step": 2510 }, { - "epoch": 0.14, - "learning_rate": 1.7220278749514134e-05, - "loss": 0.1246, + "epoch": 0.28, + "learning_rate": 1.7220124389160374e-05, + "loss": 0.0961, "step": 2515 }, { - "epoch": 0.14, - "learning_rate": 1.721472597034816e-05, - "loss": 0.1256, + "epoch": 0.28, + "learning_rate": 1.7214571301643713e-05, + "loss": 0.1323, "step": 2520 }, { - "epoch": 0.14, - "learning_rate": 1.7209173191182188e-05, - "loss": 0.1685, + "epoch": 0.28, + "learning_rate": 1.7209018214127056e-05, + "loss": 0.0683, "step": 2525 }, { - "epoch": 0.14, - "learning_rate": 1.7203620412016216e-05, - "loss": 0.0854, + "epoch": 0.28, + "learning_rate": 1.7203465126610395e-05, + "loss": 0.0736, "step": 2530 }, { - "epoch": 0.14, - "learning_rate": 1.7198067632850245e-05, - "loss": 0.1254, + "epoch": 0.28, + "learning_rate": 1.7197912039093738e-05, + "loss": 0.0852, "step": 2535 }, { - "epoch": 0.14, - "learning_rate": 1.719251485368427e-05, - "loss": 0.1322, + "epoch": 0.28, + "learning_rate": 1.7192358951577077e-05, + "loss": 0.1242, "step": 2540 }, { - "epoch": 0.14, - "learning_rate": 1.71869620745183e-05, - "loss": 0.2009, + "epoch": 0.28, + "learning_rate": 1.718680586406042e-05, + "loss": 0.207, "step": 2545 }, { - "epoch": 0.14, - "learning_rate": 1.7181409295352324e-05, - "loss": 0.3264, + "epoch": 0.28, + "learning_rate": 1.718125277654376e-05, + "loss": 0.396, "step": 2550 }, { - "epoch": 0.14, - "learning_rate": 1.7175856516186353e-05, - "loss": 0.1524, + "epoch": 0.28, + "learning_rate": 1.7175699689027102e-05, + "loss": 0.0785, "step": 2555 }, { - "epoch": 0.14, - "learning_rate": 1.7170303737020378e-05, - "loss": 0.1429, + "epoch": 0.28, + "learning_rate": 1.717014660151044e-05, + "loss": 0.1103, "step": 2560 }, { - "epoch": 0.14, - "learning_rate": 1.7164750957854407e-05, - "loss": 0.1513, + "epoch": 0.28, + "learning_rate": 1.716459351399378e-05, + "loss": 0.084, "step": 2565 }, { - "epoch": 0.14, - "learning_rate": 1.7159198178688435e-05, - "loss": 0.1638, + "epoch": 0.29, + "learning_rate": 1.7159040426477124e-05, + "loss": 0.1304, "step": 2570 }, { - "epoch": 0.14, - "learning_rate": 1.7153645399522464e-05, - "loss": 0.1514, + "epoch": 0.29, + "learning_rate": 1.7153487338960463e-05, + "loss": 0.0576, "step": 2575 }, { - "epoch": 0.14, - "learning_rate": 1.714809262035649e-05, - "loss": 0.0963, + "epoch": 0.29, + "learning_rate": 1.7147934251443806e-05, + "loss": 0.0594, "step": 2580 }, { - "epoch": 0.14, - "learning_rate": 1.7142539841190518e-05, - "loss": 0.1343, + "epoch": 0.29, + "learning_rate": 1.7142381163927145e-05, + "loss": 0.0975, "step": 2585 }, { - "epoch": 0.14, - "learning_rate": 1.7136987062024543e-05, - "loss": 0.0946, + "epoch": 0.29, + "learning_rate": 1.7136828076410485e-05, + "loss": 0.1714, "step": 2590 }, { - "epoch": 0.14, - "learning_rate": 1.7131434282858572e-05, - "loss": 0.1845, + "epoch": 0.29, + "learning_rate": 1.7131274988893827e-05, + "loss": 0.1486, "step": 2595 }, { - "epoch": 0.14, - "learning_rate": 1.7125881503692597e-05, - "loss": 0.5798, + "epoch": 0.29, + "learning_rate": 1.7125721901377167e-05, + "loss": 0.3571, "step": 2600 }, { - "epoch": 0.14, - "learning_rate": 1.7120328724526626e-05, - "loss": 0.2439, + "epoch": 0.29, + "learning_rate": 1.7120168813860506e-05, + "loss": 0.1006, "step": 2605 }, { - "epoch": 0.14, - "learning_rate": 1.7114775945360654e-05, - "loss": 0.1155, + "epoch": 0.29, + "learning_rate": 1.711461572634385e-05, + "loss": 0.1108, "step": 2610 }, { - "epoch": 0.15, - "learning_rate": 1.7109223166194683e-05, - "loss": 0.1356, + "epoch": 0.29, + "learning_rate": 1.7109062638827188e-05, + "loss": 0.1039, "step": 2615 }, { - "epoch": 0.15, - "learning_rate": 1.710367038702871e-05, - "loss": 0.2003, + "epoch": 0.29, + "learning_rate": 1.710350955131053e-05, + "loss": 0.117, "step": 2620 }, { - "epoch": 0.15, - "learning_rate": 1.7098117607862737e-05, - "loss": 0.0968, + "epoch": 0.29, + "learning_rate": 1.709795646379387e-05, + "loss": 0.1117, "step": 2625 }, { - "epoch": 0.15, - "learning_rate": 1.7092564828696765e-05, - "loss": 0.0573, + "epoch": 0.29, + "learning_rate": 1.709240337627721e-05, + "loss": 0.0795, "step": 2630 }, { - "epoch": 0.15, - "learning_rate": 1.708701204953079e-05, - "loss": 0.1765, + "epoch": 0.29, + "learning_rate": 1.7086850288760552e-05, + "loss": 0.0756, "step": 2635 }, { - "epoch": 0.15, - "learning_rate": 1.708145927036482e-05, - "loss": 0.1691, + "epoch": 0.29, + "learning_rate": 1.7081297201243895e-05, + "loss": 0.1152, "step": 2640 }, { - "epoch": 0.15, - "learning_rate": 1.7075906491198845e-05, - "loss": 0.2128, + "epoch": 0.29, + "learning_rate": 1.707574411372723e-05, + "loss": 0.2112, "step": 2645 }, { - "epoch": 0.15, - "learning_rate": 1.7070353712032873e-05, - "loss": 0.2991, + "epoch": 0.29, + "learning_rate": 1.7070191026210574e-05, + "loss": 0.3646, "step": 2650 }, { - "epoch": 0.15, - "learning_rate": 1.7064800932866902e-05, - "loss": 0.1996, + "epoch": 0.29, + "learning_rate": 1.7064637938693917e-05, + "loss": 0.078, "step": 2655 }, { - "epoch": 0.15, - "learning_rate": 1.705924815370093e-05, - "loss": 0.1204, + "epoch": 0.3, + "learning_rate": 1.7059084851177256e-05, + "loss": 0.0902, "step": 2660 }, { - "epoch": 0.15, - "learning_rate": 1.7053695374534956e-05, - "loss": 0.1239, + "epoch": 0.3, + "learning_rate": 1.7053531763660595e-05, + "loss": 0.0971, "step": 2665 }, { - "epoch": 0.15, - "learning_rate": 1.7048142595368984e-05, - "loss": 0.1541, + "epoch": 0.3, + "learning_rate": 1.7047978676143938e-05, + "loss": 0.1079, "step": 2670 }, { - "epoch": 0.15, - "learning_rate": 1.704258981620301e-05, - "loss": 0.19, + "epoch": 0.3, + "learning_rate": 1.7042425588627278e-05, + "loss": 0.061, "step": 2675 }, { - "epoch": 0.15, - "learning_rate": 1.7037037037037038e-05, - "loss": 0.0699, + "epoch": 0.3, + "learning_rate": 1.703687250111062e-05, + "loss": 0.0765, "step": 2680 }, { - "epoch": 0.15, - "learning_rate": 1.7031484257871064e-05, - "loss": 0.082, + "epoch": 0.3, + "learning_rate": 1.703131941359396e-05, + "loss": 0.1012, "step": 2685 }, { - "epoch": 0.15, - "learning_rate": 1.7025931478705092e-05, - "loss": 0.1937, + "epoch": 0.3, + "learning_rate": 1.70257663260773e-05, + "loss": 0.0899, "step": 2690 }, { - "epoch": 0.15, - "learning_rate": 1.702037869953912e-05, - "loss": 0.2674, + "epoch": 0.3, + "learning_rate": 1.7020213238560642e-05, + "loss": 0.1788, "step": 2695 }, { - "epoch": 0.15, - "learning_rate": 1.701482592037315e-05, - "loss": 0.6053, + "epoch": 0.3, + "learning_rate": 1.701466015104398e-05, + "loss": 0.2892, "step": 2700 }, { - "epoch": 0.15, - "learning_rate": 1.7009273141207178e-05, - "loss": 0.1209, + "epoch": 0.3, + "learning_rate": 1.7009107063527324e-05, + "loss": 0.0653, "step": 2705 }, { - "epoch": 0.15, - "learning_rate": 1.7003720362041203e-05, - "loss": 0.1639, + "epoch": 0.3, + "learning_rate": 1.7003553976010663e-05, + "loss": 0.0878, "step": 2710 }, { - "epoch": 0.15, - "learning_rate": 1.6998167582875232e-05, - "loss": 0.1417, + "epoch": 0.3, + "learning_rate": 1.6998000888494003e-05, + "loss": 0.1398, "step": 2715 }, { - "epoch": 0.15, - "learning_rate": 1.6992614803709257e-05, - "loss": 0.1252, + "epoch": 0.3, + "learning_rate": 1.6992447800977345e-05, + "loss": 0.1062, "step": 2720 }, { - "epoch": 0.15, - "learning_rate": 1.6987062024543286e-05, - "loss": 0.1953, + "epoch": 0.3, + "learning_rate": 1.6986894713460685e-05, + "loss": 0.0806, "step": 2725 }, { - "epoch": 0.15, - "learning_rate": 1.698150924537731e-05, - "loss": 0.2, + "epoch": 0.3, + "learning_rate": 1.6981341625944024e-05, + "loss": 0.0911, "step": 2730 }, { - "epoch": 0.15, - "learning_rate": 1.697595646621134e-05, - "loss": 0.1698, + "epoch": 0.3, + "learning_rate": 1.6975788538427367e-05, + "loss": 0.082, "step": 2735 }, { - "epoch": 0.15, - "learning_rate": 1.697040368704537e-05, - "loss": 0.2235, + "epoch": 0.3, + "learning_rate": 1.697023545091071e-05, + "loss": 0.1154, "step": 2740 }, { - "epoch": 0.15, - "learning_rate": 1.6964850907879397e-05, - "loss": 0.1965, + "epoch": 0.3, + "learning_rate": 1.696468236339405e-05, + "loss": 0.1959, "step": 2745 }, { - "epoch": 0.15, - "learning_rate": 1.6959298128713422e-05, - "loss": 0.5831, + "epoch": 0.31, + "learning_rate": 1.695912927587739e-05, + "loss": 0.367, "step": 2750 }, { - "epoch": 0.15, - "learning_rate": 1.695374534954745e-05, - "loss": 0.1557, + "epoch": 0.31, + "learning_rate": 1.695357618836073e-05, + "loss": 0.0805, "step": 2755 }, { - "epoch": 0.15, - "learning_rate": 1.6948192570381476e-05, - "loss": 0.1599, + "epoch": 0.31, + "learning_rate": 1.694802310084407e-05, + "loss": 0.1069, "step": 2760 }, { - "epoch": 0.15, - "learning_rate": 1.6942639791215505e-05, - "loss": 0.157, + "epoch": 0.31, + "learning_rate": 1.6942470013327413e-05, + "loss": 0.1016, "step": 2765 }, { - "epoch": 0.15, - "learning_rate": 1.693708701204953e-05, - "loss": 0.1782, + "epoch": 0.31, + "learning_rate": 1.6936916925810753e-05, + "loss": 0.1137, "step": 2770 }, { - "epoch": 0.15, - "learning_rate": 1.693153423288356e-05, - "loss": 0.1515, + "epoch": 0.31, + "learning_rate": 1.6931363838294092e-05, + "loss": 0.0935, "step": 2775 }, { - "epoch": 0.15, - "learning_rate": 1.6925981453717587e-05, - "loss": 0.1476, + "epoch": 0.31, + "learning_rate": 1.6925810750777435e-05, + "loss": 0.0761, "step": 2780 }, { - "epoch": 0.15, - "learning_rate": 1.6920428674551616e-05, - "loss": 0.1089, + "epoch": 0.31, + "learning_rate": 1.6920257663260774e-05, + "loss": 0.0943, "step": 2785 }, { - "epoch": 0.15, - "learning_rate": 1.691487589538564e-05, - "loss": 0.1913, + "epoch": 0.31, + "learning_rate": 1.6914704575744114e-05, + "loss": 0.1072, "step": 2790 }, { - "epoch": 0.16, - "learning_rate": 1.690932311621967e-05, - "loss": 0.2587, + "epoch": 0.31, + "learning_rate": 1.6909151488227456e-05, + "loss": 0.2146, "step": 2795 }, { - "epoch": 0.16, - "learning_rate": 1.69037703370537e-05, - "loss": 0.4758, + "epoch": 0.31, + "learning_rate": 1.6903598400710796e-05, + "loss": 0.3725, "step": 2800 }, { - "epoch": 0.16, - "learning_rate": 1.6898217557887724e-05, - "loss": 0.1408, + "epoch": 0.31, + "learning_rate": 1.689804531319414e-05, + "loss": 0.092, "step": 2805 }, { - "epoch": 0.16, - "learning_rate": 1.6892664778721753e-05, - "loss": 0.1284, + "epoch": 0.31, + "learning_rate": 1.6892492225677478e-05, + "loss": 0.0832, "step": 2810 }, { - "epoch": 0.16, - "learning_rate": 1.6887111999555778e-05, - "loss": 0.1241, + "epoch": 0.31, + "learning_rate": 1.6886939138160817e-05, + "loss": 0.1009, "step": 2815 }, { - "epoch": 0.16, - "learning_rate": 1.6881559220389806e-05, - "loss": 0.1625, + "epoch": 0.31, + "learning_rate": 1.688138605064416e-05, + "loss": 0.0979, "step": 2820 }, { - "epoch": 0.16, - "learning_rate": 1.687600644122383e-05, - "loss": 0.157, + "epoch": 0.31, + "learning_rate": 1.68758329631275e-05, + "loss": 0.0872, "step": 2825 }, { - "epoch": 0.16, - "learning_rate": 1.687045366205786e-05, - "loss": 0.1395, + "epoch": 0.31, + "learning_rate": 1.687027987561084e-05, + "loss": 0.1036, "step": 2830 }, { - "epoch": 0.16, - "learning_rate": 1.686490088289189e-05, - "loss": 0.0849, + "epoch": 0.31, + "learning_rate": 1.686472678809418e-05, + "loss": 0.0925, "step": 2835 }, { - "epoch": 0.16, - "learning_rate": 1.6859348103725918e-05, - "loss": 0.1335, + "epoch": 0.32, + "learning_rate": 1.685917370057752e-05, + "loss": 0.145, "step": 2840 }, { - "epoch": 0.16, - "learning_rate": 1.6853795324559943e-05, - "loss": 0.2904, + "epoch": 0.32, + "learning_rate": 1.6853620613060864e-05, + "loss": 0.1965, "step": 2845 }, { - "epoch": 0.16, - "learning_rate": 1.684824254539397e-05, - "loss": 0.486, + "epoch": 0.32, + "learning_rate": 1.6848067525544206e-05, + "loss": 0.3851, "step": 2850 }, { - "epoch": 0.16, - "learning_rate": 1.6842689766227997e-05, - "loss": 0.1444, + "epoch": 0.32, + "learning_rate": 1.6842514438027546e-05, + "loss": 0.1303, "step": 2855 }, { - "epoch": 0.16, - "learning_rate": 1.6837136987062025e-05, - "loss": 0.1226, + "epoch": 0.32, + "learning_rate": 1.6836961350510885e-05, + "loss": 0.0944, "step": 2860 }, { - "epoch": 0.16, - "learning_rate": 1.683158420789605e-05, - "loss": 0.1376, + "epoch": 0.32, + "learning_rate": 1.6831408262994228e-05, + "loss": 0.1084, "step": 2865 }, { - "epoch": 0.16, - "learning_rate": 1.682603142873008e-05, - "loss": 0.1411, + "epoch": 0.32, + "learning_rate": 1.6825855175477567e-05, + "loss": 0.1098, "step": 2870 }, { - "epoch": 0.16, - "learning_rate": 1.6820478649564108e-05, - "loss": 0.1352, + "epoch": 0.32, + "learning_rate": 1.6820302087960907e-05, + "loss": 0.1134, "step": 2875 }, { - "epoch": 0.16, - "learning_rate": 1.6814925870398137e-05, - "loss": 0.085, + "epoch": 0.32, + "learning_rate": 1.681474900044425e-05, + "loss": 0.0863, "step": 2880 }, { - "epoch": 0.16, - "learning_rate": 1.6809373091232165e-05, - "loss": 0.1289, + "epoch": 0.32, + "learning_rate": 1.680919591292759e-05, + "loss": 0.0932, "step": 2885 }, { - "epoch": 0.16, - "learning_rate": 1.680382031206619e-05, - "loss": 0.1806, + "epoch": 0.32, + "learning_rate": 1.680364282541093e-05, + "loss": 0.1401, "step": 2890 }, { - "epoch": 0.16, - "learning_rate": 1.679826753290022e-05, - "loss": 0.2307, + "epoch": 0.32, + "learning_rate": 1.679808973789427e-05, + "loss": 0.1387, "step": 2895 }, { - "epoch": 0.16, - "learning_rate": 1.6792714753734244e-05, - "loss": 0.3548, + "epoch": 0.32, + "learning_rate": 1.679253665037761e-05, + "loss": 0.2855, "step": 2900 }, { - "epoch": 0.16, - "learning_rate": 1.6787161974568273e-05, - "loss": 0.1403, + "epoch": 0.32, + "learning_rate": 1.6786983562860953e-05, + "loss": 0.0765, "step": 2905 }, { - "epoch": 0.16, - "learning_rate": 1.6781609195402298e-05, - "loss": 0.1414, + "epoch": 0.32, + "learning_rate": 1.6781430475344292e-05, + "loss": 0.1152, "step": 2910 }, { - "epoch": 0.16, - "learning_rate": 1.6776056416236327e-05, - "loss": 0.2085, + "epoch": 0.32, + "learning_rate": 1.677587738782763e-05, + "loss": 0.0868, "step": 2915 }, { - "epoch": 0.16, - "learning_rate": 1.6770503637070356e-05, - "loss": 0.2353, + "epoch": 0.32, + "learning_rate": 1.6770324300310974e-05, + "loss": 0.0992, "step": 2920 }, { - "epoch": 0.16, - "learning_rate": 1.6764950857904384e-05, - "loss": 0.1097, + "epoch": 0.32, + "learning_rate": 1.6764771212794314e-05, + "loss": 0.091, "step": 2925 }, { - "epoch": 0.16, - "learning_rate": 1.675939807873841e-05, - "loss": 0.0473, + "epoch": 0.33, + "learning_rate": 1.6759218125277657e-05, + "loss": 0.0412, "step": 2930 }, { - "epoch": 0.16, - "learning_rate": 1.6753845299572438e-05, - "loss": 0.1121, + "epoch": 0.33, + "learning_rate": 1.6753665037760996e-05, + "loss": 0.0889, "step": 2935 }, { - "epoch": 0.16, - "learning_rate": 1.6748292520406463e-05, - "loss": 0.1378, + "epoch": 0.33, + "learning_rate": 1.6748111950244335e-05, + "loss": 0.1182, "step": 2940 }, { - "epoch": 0.16, - "learning_rate": 1.6742739741240492e-05, - "loss": 0.2852, + "epoch": 0.33, + "learning_rate": 1.6742558862727678e-05, + "loss": 0.1759, "step": 2945 }, { - "epoch": 0.16, - "learning_rate": 1.6737186962074517e-05, - "loss": 0.4812, + "epoch": 0.33, + "learning_rate": 1.673700577521102e-05, + "loss": 0.3003, "step": 2950 }, { - "epoch": 0.16, - "learning_rate": 1.6731634182908546e-05, - "loss": 0.1464, + "epoch": 0.33, + "learning_rate": 1.6731452687694357e-05, + "loss": 0.1007, "step": 2955 }, { - "epoch": 0.16, - "learning_rate": 1.6726081403742575e-05, - "loss": 0.1142, + "epoch": 0.33, + "learning_rate": 1.67258996001777e-05, + "loss": 0.0859, "step": 2960 }, { - "epoch": 0.16, - "learning_rate": 1.6720528624576603e-05, - "loss": 0.2094, + "epoch": 0.33, + "learning_rate": 1.6720346512661042e-05, + "loss": 0.0887, "step": 2965 }, { - "epoch": 0.16, - "learning_rate": 1.6714975845410632e-05, - "loss": 0.1422, + "epoch": 0.33, + "learning_rate": 1.671479342514438e-05, + "loss": 0.1244, "step": 2970 }, { - "epoch": 0.17, - "learning_rate": 1.6709423066244657e-05, - "loss": 0.1275, + "epoch": 0.33, + "learning_rate": 1.6709240337627724e-05, + "loss": 0.0652, "step": 2975 }, { - "epoch": 0.17, - "learning_rate": 1.6703870287078686e-05, - "loss": 0.0794, + "epoch": 0.33, + "learning_rate": 1.6703687250111064e-05, + "loss": 0.0617, "step": 2980 }, { - "epoch": 0.17, - "learning_rate": 1.669831750791271e-05, - "loss": 0.1196, + "epoch": 0.33, + "learning_rate": 1.6698134162594403e-05, + "loss": 0.1101, "step": 2985 }, { - "epoch": 0.17, - "learning_rate": 1.669276472874674e-05, - "loss": 0.177, + "epoch": 0.33, + "learning_rate": 1.6692581075077746e-05, + "loss": 0.1038, "step": 2990 }, { - "epoch": 0.17, - "learning_rate": 1.6687211949580765e-05, - "loss": 0.2154, + "epoch": 0.33, + "learning_rate": 1.6687027987561085e-05, + "loss": 0.1654, "step": 2995 }, { - "epoch": 0.17, - "learning_rate": 1.6681659170414794e-05, - "loss": 0.429, + "epoch": 0.33, + "learning_rate": 1.6681474900044425e-05, + "loss": 0.2911, "step": 3000 }, { - "epoch": 0.17, - "learning_rate": 1.6676106391248822e-05, - "loss": 0.1085, + "epoch": 0.33, + "learning_rate": 1.6675921812527767e-05, + "loss": 0.077, "step": 3005 }, { - "epoch": 0.17, - "learning_rate": 1.667055361208285e-05, - "loss": 0.1681, + "epoch": 0.33, + "learning_rate": 1.6670368725011107e-05, + "loss": 0.0814, "step": 3010 }, { - "epoch": 0.17, - "learning_rate": 1.6665000832916876e-05, - "loss": 0.1425, + "epoch": 0.33, + "learning_rate": 1.666481563749445e-05, + "loss": 0.1077, "step": 3015 }, { - "epoch": 0.17, - "learning_rate": 1.6659448053750905e-05, - "loss": 0.1425, + "epoch": 0.34, + "learning_rate": 1.665926254997779e-05, + "loss": 0.1065, "step": 3020 }, { - "epoch": 0.17, - "learning_rate": 1.665389527458493e-05, - "loss": 0.0753, + "epoch": 0.34, + "learning_rate": 1.6653709462461128e-05, + "loss": 0.0641, "step": 3025 }, { - "epoch": 0.17, - "learning_rate": 1.664834249541896e-05, - "loss": 0.1961, + "epoch": 0.34, + "learning_rate": 1.664815637494447e-05, + "loss": 0.0768, "step": 3030 }, { - "epoch": 0.17, - "learning_rate": 1.6642789716252984e-05, - "loss": 0.1514, + "epoch": 0.34, + "learning_rate": 1.664260328742781e-05, + "loss": 0.096, "step": 3035 }, { - "epoch": 0.17, - "learning_rate": 1.6637236937087012e-05, - "loss": 0.1701, + "epoch": 0.34, + "learning_rate": 1.663705019991115e-05, + "loss": 0.0943, "step": 3040 }, { - "epoch": 0.17, - "learning_rate": 1.663168415792104e-05, - "loss": 0.2889, + "epoch": 0.34, + "learning_rate": 1.6631497112394492e-05, + "loss": 0.1964, "step": 3045 }, { - "epoch": 0.17, - "learning_rate": 1.662613137875507e-05, - "loss": 0.4184, + "epoch": 0.34, + "learning_rate": 1.6625944024877835e-05, + "loss": 0.2383, "step": 3050 }, { - "epoch": 0.17, - "learning_rate": 1.6620578599589095e-05, - "loss": 0.1446, + "epoch": 0.34, + "learning_rate": 1.6620390937361175e-05, + "loss": 0.1128, "step": 3055 }, { - "epoch": 0.17, - "learning_rate": 1.6615025820423124e-05, - "loss": 0.1463, + "epoch": 0.34, + "learning_rate": 1.6614837849844514e-05, + "loss": 0.0922, "step": 3060 }, { - "epoch": 0.17, - "learning_rate": 1.6609473041257152e-05, - "loss": 0.2113, + "epoch": 0.34, + "learning_rate": 1.6609284762327857e-05, + "loss": 0.0897, "step": 3065 }, { - "epoch": 0.17, - "learning_rate": 1.6603920262091178e-05, - "loss": 0.1345, + "epoch": 0.34, + "learning_rate": 1.6603731674811196e-05, + "loss": 0.1518, "step": 3070 }, { - "epoch": 0.17, - "learning_rate": 1.6598367482925206e-05, - "loss": 0.0884, + "epoch": 0.34, + "learning_rate": 1.659817858729454e-05, + "loss": 0.0945, "step": 3075 }, { - "epoch": 0.17, - "learning_rate": 1.659281470375923e-05, - "loss": 0.0528, + "epoch": 0.34, + "learning_rate": 1.6592625499777878e-05, + "loss": 0.0477, "step": 3080 }, { - "epoch": 0.17, - "learning_rate": 1.658726192459326e-05, - "loss": 0.1216, + "epoch": 0.34, + "learning_rate": 1.6587072412261218e-05, + "loss": 0.112, "step": 3085 }, { - "epoch": 0.17, - "learning_rate": 1.658170914542729e-05, - "loss": 0.0975, + "epoch": 0.34, + "learning_rate": 1.658151932474456e-05, + "loss": 0.1762, "step": 3090 }, { - "epoch": 0.17, - "learning_rate": 1.6576156366261314e-05, - "loss": 0.2197, + "epoch": 0.34, + "learning_rate": 1.65759662372279e-05, + "loss": 0.1103, "step": 3095 }, { - "epoch": 0.17, - "learning_rate": 1.6570603587095343e-05, - "loss": 0.461, + "epoch": 0.34, + "learning_rate": 1.657041314971124e-05, + "loss": 0.3118, "step": 3100 }, { - "epoch": 0.17, - "learning_rate": 1.656505080792937e-05, - "loss": 0.1514, + "epoch": 0.34, + "learning_rate": 1.6564860062194582e-05, + "loss": 0.1042, "step": 3105 }, { - "epoch": 0.17, - "learning_rate": 1.6559498028763397e-05, - "loss": 0.1547, + "epoch": 0.35, + "learning_rate": 1.655930697467792e-05, + "loss": 0.0835, "step": 3110 }, { - "epoch": 0.17, - "learning_rate": 1.6553945249597425e-05, - "loss": 0.1535, + "epoch": 0.35, + "learning_rate": 1.6553753887161264e-05, + "loss": 0.0869, "step": 3115 }, { - "epoch": 0.17, - "learning_rate": 1.654839247043145e-05, - "loss": 0.1783, + "epoch": 0.35, + "learning_rate": 1.6548200799644603e-05, + "loss": 0.0982, "step": 3120 }, { - "epoch": 0.17, - "learning_rate": 1.654283969126548e-05, - "loss": 0.1521, + "epoch": 0.35, + "learning_rate": 1.6542647712127943e-05, + "loss": 0.1103, "step": 3125 }, { - "epoch": 0.17, - "learning_rate": 1.6537286912099508e-05, - "loss": 0.099, + "epoch": 0.35, + "learning_rate": 1.6537094624611285e-05, + "loss": 0.0477, "step": 3130 }, { - "epoch": 0.17, - "learning_rate": 1.6531734132933533e-05, - "loss": 0.1591, + "epoch": 0.35, + "learning_rate": 1.6531541537094625e-05, + "loss": 0.0709, "step": 3135 }, { - "epoch": 0.17, - "learning_rate": 1.652618135376756e-05, - "loss": 0.192, + "epoch": 0.35, + "learning_rate": 1.6525988449577968e-05, + "loss": 0.0971, "step": 3140 }, { - "epoch": 0.17, - "learning_rate": 1.652062857460159e-05, - "loss": 0.2343, + "epoch": 0.35, + "learning_rate": 1.6520435362061307e-05, + "loss": 0.1319, "step": 3145 }, { - "epoch": 0.17, - "learning_rate": 1.651507579543562e-05, - "loss": 0.349, + "epoch": 0.35, + "learning_rate": 1.6514882274544646e-05, + "loss": 0.2376, "step": 3150 }, { - "epoch": 0.18, - "learning_rate": 1.6509523016269644e-05, - "loss": 0.1041, + "epoch": 0.35, + "learning_rate": 1.650932918702799e-05, + "loss": 0.0834, "step": 3155 }, { - "epoch": 0.18, - "learning_rate": 1.6503970237103673e-05, - "loss": 0.1231, + "epoch": 0.35, + "learning_rate": 1.6503776099511332e-05, + "loss": 0.0953, "step": 3160 }, { - "epoch": 0.18, - "learning_rate": 1.6498417457937698e-05, - "loss": 0.1092, + "epoch": 0.35, + "learning_rate": 1.649822301199467e-05, + "loss": 0.1127, "step": 3165 }, { - "epoch": 0.18, - "learning_rate": 1.6492864678771727e-05, - "loss": 0.1486, + "epoch": 0.35, + "learning_rate": 1.649266992447801e-05, + "loss": 0.1344, "step": 3170 }, { - "epoch": 0.18, - "learning_rate": 1.6487311899605752e-05, - "loss": 0.151, + "epoch": 0.35, + "learning_rate": 1.6487116836961353e-05, + "loss": 0.0798, "step": 3175 }, { - "epoch": 0.18, - "learning_rate": 1.648175912043978e-05, - "loss": 0.1008, + "epoch": 0.35, + "learning_rate": 1.6481563749444693e-05, + "loss": 0.0569, "step": 3180 }, { - "epoch": 0.18, - "learning_rate": 1.647620634127381e-05, - "loss": 0.1591, + "epoch": 0.35, + "learning_rate": 1.6476010661928032e-05, + "loss": 0.0954, "step": 3185 }, { - "epoch": 0.18, - "learning_rate": 1.6470653562107838e-05, - "loss": 0.1719, + "epoch": 0.35, + "learning_rate": 1.6470457574411375e-05, + "loss": 0.0844, "step": 3190 }, { - "epoch": 0.18, - "learning_rate": 1.6465100782941863e-05, - "loss": 0.2631, + "epoch": 0.35, + "learning_rate": 1.6464904486894714e-05, + "loss": 0.1709, "step": 3195 }, { - "epoch": 0.18, - "learning_rate": 1.6459548003775892e-05, - "loss": 0.5223, + "epoch": 0.36, + "learning_rate": 1.6459351399378057e-05, + "loss": 0.3021, "step": 3200 }, { - "epoch": 0.18, - "learning_rate": 1.6453995224609917e-05, - "loss": 0.1353, + "epoch": 0.36, + "learning_rate": 1.6453798311861396e-05, + "loss": 0.0801, "step": 3205 }, { - "epoch": 0.18, - "learning_rate": 1.6448442445443946e-05, - "loss": 0.1734, + "epoch": 0.36, + "learning_rate": 1.6448245224344736e-05, + "loss": 0.099, "step": 3210 }, { - "epoch": 0.18, - "learning_rate": 1.644288966627797e-05, - "loss": 0.1603, + "epoch": 0.36, + "learning_rate": 1.644269213682808e-05, + "loss": 0.1008, "step": 3215 }, { - "epoch": 0.18, - "learning_rate": 1.6437336887112e-05, - "loss": 0.1901, + "epoch": 0.36, + "learning_rate": 1.6437139049311418e-05, + "loss": 0.1072, "step": 3220 }, { - "epoch": 0.18, - "learning_rate": 1.6431784107946028e-05, - "loss": 0.1134, + "epoch": 0.36, + "learning_rate": 1.6431585961794757e-05, + "loss": 0.0853, "step": 3225 }, { - "epoch": 0.18, - "learning_rate": 1.6426231328780057e-05, - "loss": 0.1267, + "epoch": 0.36, + "learning_rate": 1.64260328742781e-05, + "loss": 0.0733, "step": 3230 }, { - "epoch": 0.18, - "learning_rate": 1.6420678549614086e-05, - "loss": 0.1872, + "epoch": 0.36, + "learning_rate": 1.642047978676144e-05, + "loss": 0.0471, "step": 3235 }, { - "epoch": 0.18, - "learning_rate": 1.641512577044811e-05, - "loss": 0.2061, + "epoch": 0.36, + "learning_rate": 1.6414926699244782e-05, + "loss": 0.1344, "step": 3240 }, { - "epoch": 0.18, - "learning_rate": 1.640957299128214e-05, - "loss": 0.2511, + "epoch": 0.36, + "learning_rate": 1.6409373611728125e-05, + "loss": 0.0853, "step": 3245 }, { - "epoch": 0.18, - "learning_rate": 1.640513076794936e-05, - "loss": 0.5597, + "epoch": 0.36, + "learning_rate": 1.640382052421146e-05, + "loss": 0.2943, "step": 3250 }, { - "epoch": 0.18, - "learning_rate": 1.6399577988783388e-05, - "loss": 0.114, + "epoch": 0.36, + "learning_rate": 1.6398267436694804e-05, + "loss": 0.0834, "step": 3255 }, { - "epoch": 0.18, - "learning_rate": 1.6394025209617416e-05, - "loss": 0.1155, + "epoch": 0.36, + "learning_rate": 1.6392714349178146e-05, + "loss": 0.0791, "step": 3260 }, { - "epoch": 0.18, - "learning_rate": 1.638847243045144e-05, - "loss": 0.1557, + "epoch": 0.36, + "learning_rate": 1.6387161261661486e-05, + "loss": 0.1087, "step": 3265 }, { - "epoch": 0.18, - "learning_rate": 1.638291965128547e-05, - "loss": 0.1587, + "epoch": 0.36, + "learning_rate": 1.6381608174144825e-05, + "loss": 0.1174, "step": 3270 }, { - "epoch": 0.18, - "learning_rate": 1.63773668721195e-05, - "loss": 0.1021, + "epoch": 0.36, + "learning_rate": 1.6376055086628168e-05, + "loss": 0.1023, "step": 3275 }, { - "epoch": 0.18, - "learning_rate": 1.6371814092953524e-05, - "loss": 0.0358, + "epoch": 0.36, + "learning_rate": 1.6370501999111507e-05, + "loss": 0.0464, "step": 3280 }, { - "epoch": 0.18, - "learning_rate": 1.6366261313787553e-05, - "loss": 0.1259, + "epoch": 0.36, + "learning_rate": 1.636494891159485e-05, + "loss": 0.0904, "step": 3285 }, { - "epoch": 0.18, - "learning_rate": 1.6360708534621578e-05, - "loss": 0.21, + "epoch": 0.37, + "learning_rate": 1.635939582407819e-05, + "loss": 0.1299, "step": 3290 }, { - "epoch": 0.18, - "learning_rate": 1.6355155755455607e-05, - "loss": 0.3551, + "epoch": 0.37, + "learning_rate": 1.635384273656153e-05, + "loss": 0.2206, "step": 3295 }, { - "epoch": 0.18, - "learning_rate": 1.6349602976289635e-05, - "loss": 0.3455, + "epoch": 0.37, + "learning_rate": 1.634828964904487e-05, + "loss": 0.4083, "step": 3300 }, { - "epoch": 0.18, - "learning_rate": 1.634405019712366e-05, - "loss": 0.129, + "epoch": 0.37, + "learning_rate": 1.634273656152821e-05, + "loss": 0.0842, "step": 3305 }, { - "epoch": 0.18, - "learning_rate": 1.633849741795769e-05, - "loss": 0.1622, + "epoch": 0.37, + "learning_rate": 1.633718347401155e-05, + "loss": 0.1296, "step": 3310 }, { - "epoch": 0.18, - "learning_rate": 1.6332944638791718e-05, - "loss": 0.121, + "epoch": 0.37, + "learning_rate": 1.6331630386494893e-05, + "loss": 0.0928, "step": 3315 }, { - "epoch": 0.18, - "learning_rate": 1.6327391859625743e-05, - "loss": 0.1196, + "epoch": 0.37, + "learning_rate": 1.6326077298978232e-05, + "loss": 0.1003, "step": 3320 }, { - "epoch": 0.18, - "learning_rate": 1.632183908045977e-05, - "loss": 0.1226, + "epoch": 0.37, + "learning_rate": 1.6320524211461575e-05, + "loss": 0.1134, "step": 3325 }, { - "epoch": 0.18, - "learning_rate": 1.63162863012938e-05, - "loss": 0.0993, + "epoch": 0.37, + "learning_rate": 1.6314971123944914e-05, + "loss": 0.0904, "step": 3330 }, { - "epoch": 0.19, - "learning_rate": 1.6310733522127826e-05, - "loss": 0.133, + "epoch": 0.37, + "learning_rate": 1.6309418036428254e-05, + "loss": 0.058, "step": 3335 }, { - "epoch": 0.19, - "learning_rate": 1.6305180742961854e-05, - "loss": 0.1676, + "epoch": 0.37, + "learning_rate": 1.6303864948911597e-05, + "loss": 0.1375, "step": 3340 }, { - "epoch": 0.19, - "learning_rate": 1.629962796379588e-05, - "loss": 0.1016, + "epoch": 0.37, + "learning_rate": 1.6298311861394936e-05, + "loss": 0.1815, "step": 3345 }, { - "epoch": 0.19, - "learning_rate": 1.6294075184629908e-05, - "loss": 0.4682, + "epoch": 0.37, + "learning_rate": 1.6292758773878275e-05, + "loss": 0.3636, "step": 3350 }, { - "epoch": 0.19, - "learning_rate": 1.6288522405463937e-05, - "loss": 0.1356, + "epoch": 0.37, + "learning_rate": 1.6287205686361618e-05, + "loss": 0.071, "step": 3355 }, { - "epoch": 0.19, - "learning_rate": 1.6282969626297965e-05, - "loss": 0.143, + "epoch": 0.37, + "learning_rate": 1.628165259884496e-05, + "loss": 0.1088, "step": 3360 }, { - "epoch": 0.19, - "learning_rate": 1.627741684713199e-05, - "loss": 0.1583, + "epoch": 0.37, + "learning_rate": 1.62760995113283e-05, + "loss": 0.0964, "step": 3365 }, { - "epoch": 0.19, - "learning_rate": 1.627186406796602e-05, - "loss": 0.1388, + "epoch": 0.37, + "learning_rate": 1.627054642381164e-05, + "loss": 0.0859, "step": 3370 }, { - "epoch": 0.19, - "learning_rate": 1.6266311288800045e-05, - "loss": 0.1099, + "epoch": 0.37, + "learning_rate": 1.6264993336294982e-05, + "loss": 0.0795, "step": 3375 }, { - "epoch": 0.19, - "learning_rate": 1.6260758509634073e-05, - "loss": 0.1444, + "epoch": 0.38, + "learning_rate": 1.625944024877832e-05, + "loss": 0.0648, "step": 3380 }, { - "epoch": 0.19, - "learning_rate": 1.62552057304681e-05, - "loss": 0.1263, + "epoch": 0.38, + "learning_rate": 1.6253887161261664e-05, + "loss": 0.0484, "step": 3385 }, { - "epoch": 0.19, - "learning_rate": 1.6249652951302127e-05, - "loss": 0.2207, + "epoch": 0.38, + "learning_rate": 1.6248334073745004e-05, + "loss": 0.1102, "step": 3390 }, { - "epoch": 0.19, - "learning_rate": 1.6244100172136156e-05, - "loss": 0.3049, + "epoch": 0.38, + "learning_rate": 1.6242780986228343e-05, + "loss": 0.1527, "step": 3395 }, { - "epoch": 0.19, - "learning_rate": 1.6238547392970184e-05, - "loss": 0.2892, + "epoch": 0.38, + "learning_rate": 1.6237227898711686e-05, + "loss": 0.2897, "step": 3400 }, { - "epoch": 0.19, - "learning_rate": 1.623299461380421e-05, - "loss": 0.1184, + "epoch": 0.38, + "learning_rate": 1.6231674811195025e-05, + "loss": 0.0869, "step": 3405 }, { - "epoch": 0.19, - "learning_rate": 1.6227441834638238e-05, - "loss": 0.1258, + "epoch": 0.38, + "learning_rate": 1.6226121723678368e-05, + "loss": 0.1021, "step": 3410 }, { - "epoch": 0.19, - "learning_rate": 1.6221889055472263e-05, - "loss": 0.1302, + "epoch": 0.38, + "learning_rate": 1.6220568636161707e-05, + "loss": 0.124, "step": 3415 }, { - "epoch": 0.19, - "learning_rate": 1.6216336276306292e-05, - "loss": 0.157, + "epoch": 0.38, + "learning_rate": 1.6215015548645047e-05, + "loss": 0.158, "step": 3420 }, { - "epoch": 0.19, - "learning_rate": 1.621078349714032e-05, - "loss": 0.0833, + "epoch": 0.38, + "learning_rate": 1.620946246112839e-05, + "loss": 0.0583, "step": 3425 }, { - "epoch": 0.19, - "learning_rate": 1.6205230717974346e-05, - "loss": 0.0708, + "epoch": 0.38, + "learning_rate": 1.620390937361173e-05, + "loss": 0.0587, "step": 3430 }, { - "epoch": 0.19, - "learning_rate": 1.6199677938808375e-05, - "loss": 0.1559, + "epoch": 0.38, + "learning_rate": 1.6198356286095068e-05, + "loss": 0.1364, "step": 3435 }, { - "epoch": 0.19, - "learning_rate": 1.6194125159642403e-05, - "loss": 0.2079, + "epoch": 0.38, + "learning_rate": 1.619280319857841e-05, + "loss": 0.1337, "step": 3440 }, { - "epoch": 0.19, - "learning_rate": 1.6188572380476432e-05, - "loss": 0.2505, + "epoch": 0.38, + "learning_rate": 1.618725011106175e-05, + "loss": 0.2039, "step": 3445 }, { - "epoch": 0.19, - "learning_rate": 1.6183019601310457e-05, - "loss": 0.5117, + "epoch": 0.38, + "learning_rate": 1.6181697023545093e-05, + "loss": 0.4012, "step": 3450 }, { - "epoch": 0.19, - "learning_rate": 1.6177466822144486e-05, - "loss": 0.1348, + "epoch": 0.38, + "learning_rate": 1.6176143936028433e-05, + "loss": 0.0968, "step": 3455 }, { - "epoch": 0.19, - "learning_rate": 1.617191404297851e-05, - "loss": 0.1419, + "epoch": 0.38, + "learning_rate": 1.6170590848511772e-05, + "loss": 0.0984, "step": 3460 }, { - "epoch": 0.19, - "learning_rate": 1.616636126381254e-05, - "loss": 0.1009, + "epoch": 0.38, + "learning_rate": 1.6165037760995115e-05, + "loss": 0.1022, "step": 3465 }, { - "epoch": 0.19, - "learning_rate": 1.6160808484646565e-05, - "loss": 0.1368, + "epoch": 0.39, + "learning_rate": 1.6159484673478457e-05, + "loss": 0.1275, "step": 3470 }, { - "epoch": 0.19, - "learning_rate": 1.6155255705480594e-05, - "loss": 0.164, + "epoch": 0.39, + "learning_rate": 1.6153931585961797e-05, + "loss": 0.1107, "step": 3475 }, { - "epoch": 0.19, - "learning_rate": 1.6149702926314622e-05, - "loss": 0.0461, + "epoch": 0.39, + "learning_rate": 1.6148378498445136e-05, + "loss": 0.0547, "step": 3480 }, { - "epoch": 0.19, - "learning_rate": 1.614415014714865e-05, - "loss": 0.0889, + "epoch": 0.39, + "learning_rate": 1.614282541092848e-05, + "loss": 0.0904, "step": 3485 }, { - "epoch": 0.19, - "learning_rate": 1.6138597367982676e-05, - "loss": 0.1191, + "epoch": 0.39, + "learning_rate": 1.6137272323411818e-05, + "loss": 0.0762, "step": 3490 }, { - "epoch": 0.19, - "learning_rate": 1.6133044588816705e-05, - "loss": 0.1636, + "epoch": 0.39, + "learning_rate": 1.6131719235895158e-05, + "loss": 0.177, "step": 3495 }, { - "epoch": 0.19, - "learning_rate": 1.612749180965073e-05, - "loss": 0.4733, + "epoch": 0.39, + "learning_rate": 1.61261661483785e-05, + "loss": 0.2028, "step": 3500 }, { - "epoch": 0.19, - "learning_rate": 1.612193903048476e-05, - "loss": 0.0955, + "epoch": 0.39, + "learning_rate": 1.612061306086184e-05, + "loss": 0.0763, "step": 3505 }, { - "epoch": 0.19, - "learning_rate": 1.6116386251318787e-05, - "loss": 0.0872, + "epoch": 0.39, + "learning_rate": 1.6115059973345182e-05, + "loss": 0.0998, "step": 3510 }, { - "epoch": 0.2, - "learning_rate": 1.6110833472152813e-05, - "loss": 0.1474, + "epoch": 0.39, + "learning_rate": 1.6109506885828522e-05, + "loss": 0.1065, "step": 3515 }, { - "epoch": 0.2, - "learning_rate": 1.610528069298684e-05, - "loss": 0.1734, + "epoch": 0.39, + "learning_rate": 1.610395379831186e-05, + "loss": 0.0991, "step": 3520 }, { - "epoch": 0.2, - "learning_rate": 1.609972791382087e-05, - "loss": 0.1739, + "epoch": 0.39, + "learning_rate": 1.6098400710795204e-05, + "loss": 0.0556, "step": 3525 }, { - "epoch": 0.2, - "learning_rate": 1.60941751346549e-05, - "loss": 0.0636, + "epoch": 0.39, + "learning_rate": 1.6092847623278543e-05, + "loss": 0.0796, "step": 3530 }, { - "epoch": 0.2, - "learning_rate": 1.6088622355488924e-05, - "loss": 0.1268, + "epoch": 0.39, + "learning_rate": 1.6087294535761886e-05, + "loss": 0.0829, "step": 3535 }, { - "epoch": 0.2, - "learning_rate": 1.6083069576322952e-05, - "loss": 0.2232, + "epoch": 0.39, + "learning_rate": 1.6081741448245225e-05, + "loss": 0.0953, "step": 3540 }, { - "epoch": 0.2, - "learning_rate": 1.6077516797156978e-05, - "loss": 0.1895, + "epoch": 0.39, + "learning_rate": 1.6076188360728565e-05, + "loss": 0.1856, "step": 3545 }, { - "epoch": 0.2, - "learning_rate": 1.6071964017991006e-05, - "loss": 0.2901, + "epoch": 0.39, + "learning_rate": 1.6070635273211908e-05, + "loss": 0.3222, "step": 3550 }, { - "epoch": 0.2, - "learning_rate": 1.606641123882503e-05, - "loss": 0.1022, + "epoch": 0.39, + "learning_rate": 1.606508218569525e-05, + "loss": 0.0845, "step": 3555 }, { - "epoch": 0.2, - "learning_rate": 1.606085845965906e-05, - "loss": 0.1188, + "epoch": 0.4, + "learning_rate": 1.6059529098178586e-05, + "loss": 0.0808, "step": 3560 }, { - "epoch": 0.2, - "learning_rate": 1.605530568049309e-05, - "loss": 0.1551, + "epoch": 0.4, + "learning_rate": 1.605397601066193e-05, + "loss": 0.1092, "step": 3565 }, { - "epoch": 0.2, - "learning_rate": 1.6049752901327114e-05, - "loss": 0.1427, + "epoch": 0.4, + "learning_rate": 1.6048422923145272e-05, + "loss": 0.0927, "step": 3570 }, { - "epoch": 0.2, - "learning_rate": 1.6044200122161143e-05, - "loss": 0.1746, + "epoch": 0.4, + "learning_rate": 1.604286983562861e-05, + "loss": 0.112, "step": 3575 }, { - "epoch": 0.2, - "learning_rate": 1.603864734299517e-05, - "loss": 0.0779, + "epoch": 0.4, + "learning_rate": 1.603731674811195e-05, + "loss": 0.0455, "step": 3580 }, { - "epoch": 0.2, - "learning_rate": 1.6033094563829197e-05, - "loss": 0.0834, + "epoch": 0.4, + "learning_rate": 1.6031763660595293e-05, + "loss": 0.0877, "step": 3585 }, { - "epoch": 0.2, - "learning_rate": 1.6027541784663225e-05, - "loss": 0.1319, + "epoch": 0.4, + "learning_rate": 1.6026210573078633e-05, + "loss": 0.1117, "step": 3590 }, { - "epoch": 0.2, - "learning_rate": 1.6021989005497254e-05, - "loss": 0.0821, + "epoch": 0.4, + "learning_rate": 1.6020657485561975e-05, + "loss": 0.1989, "step": 3595 }, { - "epoch": 0.2, - "learning_rate": 1.601643622633128e-05, - "loss": 0.4546, + "epoch": 0.4, + "learning_rate": 1.6015104398045315e-05, + "loss": 0.3288, "step": 3600 }, { - "epoch": 0.2, - "learning_rate": 1.6010883447165308e-05, - "loss": 0.1405, + "epoch": 0.4, + "learning_rate": 1.6009551310528654e-05, + "loss": 0.0716, "step": 3605 }, { - "epoch": 0.2, - "learning_rate": 1.6005330667999333e-05, - "loss": 0.1361, + "epoch": 0.4, + "learning_rate": 1.6003998223011997e-05, + "loss": 0.0858, "step": 3610 }, { - "epoch": 0.2, - "learning_rate": 1.5999777888833362e-05, - "loss": 0.1366, + "epoch": 0.4, + "learning_rate": 1.5998445135495336e-05, + "loss": 0.0821, "step": 3615 }, { - "epoch": 0.2, - "learning_rate": 1.599422510966739e-05, - "loss": 0.203, + "epoch": 0.4, + "learning_rate": 1.5992892047978676e-05, + "loss": 0.1144, "step": 3620 }, { - "epoch": 0.2, - "learning_rate": 1.598867233050142e-05, - "loss": 0.0534, + "epoch": 0.4, + "learning_rate": 1.598733896046202e-05, + "loss": 0.0889, "step": 3625 }, { - "epoch": 0.2, - "learning_rate": 1.5983119551335444e-05, - "loss": 0.1263, + "epoch": 0.4, + "learning_rate": 1.5981785872945358e-05, + "loss": 0.0663, "step": 3630 }, { - "epoch": 0.2, - "learning_rate": 1.5977566772169473e-05, - "loss": 0.1122, + "epoch": 0.4, + "learning_rate": 1.59762327854287e-05, + "loss": 0.1213, "step": 3635 }, { - "epoch": 0.2, - "learning_rate": 1.5972013993003498e-05, - "loss": 0.1864, + "epoch": 0.4, + "learning_rate": 1.597067969791204e-05, + "loss": 0.1056, "step": 3640 }, { - "epoch": 0.2, - "learning_rate": 1.5966461213837527e-05, - "loss": 0.3265, + "epoch": 0.4, + "learning_rate": 1.596512661039538e-05, + "loss": 0.1649, "step": 3645 }, { - "epoch": 0.2, - "learning_rate": 1.5960908434671552e-05, - "loss": 0.4151, + "epoch": 0.41, + "learning_rate": 1.5959573522878722e-05, + "loss": 0.3544, "step": 3650 }, { - "epoch": 0.2, - "learning_rate": 1.595535565550558e-05, - "loss": 0.1708, + "epoch": 0.41, + "learning_rate": 1.595402043536206e-05, + "loss": 0.0892, "step": 3655 }, { - "epoch": 0.2, - "learning_rate": 1.594980287633961e-05, - "loss": 0.1372, + "epoch": 0.41, + "learning_rate": 1.5948467347845404e-05, + "loss": 0.0998, "step": 3660 }, { - "epoch": 0.2, - "learning_rate": 1.5944250097173638e-05, - "loss": 0.168, + "epoch": 0.41, + "learning_rate": 1.5942914260328744e-05, + "loss": 0.0716, "step": 3665 }, { - "epoch": 0.2, - "learning_rate": 1.5938697318007663e-05, - "loss": 0.133, + "epoch": 0.41, + "learning_rate": 1.5937361172812086e-05, + "loss": 0.0938, "step": 3670 }, { - "epoch": 0.2, - "learning_rate": 1.5933144538841692e-05, - "loss": 0.0914, + "epoch": 0.41, + "learning_rate": 1.5931808085295426e-05, + "loss": 0.0911, "step": 3675 }, { - "epoch": 0.2, - "learning_rate": 1.5927591759675717e-05, - "loss": 0.0955, + "epoch": 0.41, + "learning_rate": 1.592625499777877e-05, + "loss": 0.0515, "step": 3680 }, { - "epoch": 0.2, - "learning_rate": 1.5922038980509746e-05, - "loss": 0.1744, + "epoch": 0.41, + "learning_rate": 1.5920701910262108e-05, + "loss": 0.0547, "step": 3685 }, { - "epoch": 0.2, - "learning_rate": 1.5916486201343774e-05, - "loss": 0.1315, + "epoch": 0.41, + "learning_rate": 1.5915148822745447e-05, + "loss": 0.1367, "step": 3690 }, { - "epoch": 0.21, - "learning_rate": 1.59109334221778e-05, - "loss": 0.2831, + "epoch": 0.41, + "learning_rate": 1.590959573522879e-05, + "loss": 0.184, "step": 3695 }, { - "epoch": 0.21, - "learning_rate": 1.590538064301183e-05, - "loss": 0.5104, + "epoch": 0.41, + "learning_rate": 1.590404264771213e-05, + "loss": 0.4413, "step": 3700 }, { - "epoch": 0.21, - "learning_rate": 1.5899827863845857e-05, - "loss": 0.1115, + "epoch": 0.41, + "learning_rate": 1.589848956019547e-05, + "loss": 0.079, "step": 3705 }, { - "epoch": 0.21, - "learning_rate": 1.5894275084679886e-05, - "loss": 0.1422, + "epoch": 0.41, + "learning_rate": 1.589293647267881e-05, + "loss": 0.0785, "step": 3710 }, { - "epoch": 0.21, - "learning_rate": 1.588872230551391e-05, - "loss": 0.1812, + "epoch": 0.41, + "learning_rate": 1.588738338516215e-05, + "loss": 0.079, "step": 3715 }, { - "epoch": 0.21, - "learning_rate": 1.588316952634794e-05, - "loss": 0.1937, + "epoch": 0.41, + "learning_rate": 1.5881830297645494e-05, + "loss": 0.1071, "step": 3720 }, { - "epoch": 0.21, - "learning_rate": 1.5877616747181965e-05, - "loss": 0.0993, + "epoch": 0.41, + "learning_rate": 1.5876277210128833e-05, + "loss": 0.0562, "step": 3725 }, { - "epoch": 0.21, - "learning_rate": 1.5872063968015993e-05, - "loss": 0.0751, + "epoch": 0.41, + "learning_rate": 1.5870724122612172e-05, + "loss": 0.0541, "step": 3730 }, { - "epoch": 0.21, - "learning_rate": 1.586651118885002e-05, - "loss": 0.1206, + "epoch": 0.41, + "learning_rate": 1.5865171035095515e-05, + "loss": 0.0508, "step": 3735 }, { - "epoch": 0.21, - "learning_rate": 1.5860958409684047e-05, - "loss": 0.1861, + "epoch": 0.42, + "learning_rate": 1.5859617947578854e-05, + "loss": 0.087, "step": 3740 }, { - "epoch": 0.21, - "learning_rate": 1.5855405630518076e-05, - "loss": 0.2238, + "epoch": 0.42, + "learning_rate": 1.5854064860062194e-05, + "loss": 0.1332, "step": 3745 }, { - "epoch": 0.21, - "learning_rate": 1.5849852851352105e-05, - "loss": 0.4149, + "epoch": 0.42, + "learning_rate": 1.5848511772545537e-05, + "loss": 0.3807, "step": 3750 }, { - "epoch": 0.21, - "learning_rate": 1.584430007218613e-05, - "loss": 0.1529, + "epoch": 0.42, + "learning_rate": 1.5842958685028876e-05, + "loss": 0.0847, "step": 3755 }, { - "epoch": 0.21, - "learning_rate": 1.583874729302016e-05, - "loss": 0.1299, + "epoch": 0.42, + "learning_rate": 1.583740559751222e-05, + "loss": 0.1234, "step": 3760 }, { - "epoch": 0.21, - "learning_rate": 1.5833194513854184e-05, - "loss": 0.1473, + "epoch": 0.42, + "learning_rate": 1.5831852509995558e-05, + "loss": 0.1025, "step": 3765 }, { - "epoch": 0.21, - "learning_rate": 1.5827641734688212e-05, - "loss": 0.1844, + "epoch": 0.42, + "learning_rate": 1.5826299422478897e-05, + "loss": 0.104, "step": 3770 }, { - "epoch": 0.21, - "learning_rate": 1.582208895552224e-05, - "loss": 0.0765, + "epoch": 0.42, + "learning_rate": 1.582074633496224e-05, + "loss": 0.1179, "step": 3775 }, { - "epoch": 0.21, - "learning_rate": 1.5816536176356266e-05, - "loss": 0.1304, + "epoch": 0.42, + "learning_rate": 1.5815193247445583e-05, + "loss": 0.0519, "step": 3780 }, { - "epoch": 0.21, - "learning_rate": 1.5810983397190295e-05, - "loss": 0.2491, + "epoch": 0.42, + "learning_rate": 1.5809640159928922e-05, + "loss": 0.0564, "step": 3785 }, { - "epoch": 0.21, - "learning_rate": 1.5805430618024324e-05, - "loss": 0.1711, + "epoch": 0.42, + "learning_rate": 1.580408707241226e-05, + "loss": 0.0819, "step": 3790 }, { - "epoch": 0.21, - "learning_rate": 1.5799877838858352e-05, - "loss": 0.2845, + "epoch": 0.42, + "learning_rate": 1.5798533984895604e-05, + "loss": 0.2076, "step": 3795 }, { - "epoch": 0.21, - "learning_rate": 1.5794325059692378e-05, - "loss": 0.3614, + "epoch": 0.42, + "learning_rate": 1.5792980897378944e-05, + "loss": 0.4112, "step": 3800 }, { - "epoch": 0.21, - "learning_rate": 1.5788772280526406e-05, - "loss": 0.1268, + "epoch": 0.42, + "learning_rate": 1.5787427809862287e-05, + "loss": 0.101, "step": 3805 }, { - "epoch": 0.21, - "learning_rate": 1.578321950136043e-05, - "loss": 0.1237, + "epoch": 0.42, + "learning_rate": 1.5781874722345626e-05, + "loss": 0.0948, "step": 3810 }, { - "epoch": 0.21, - "learning_rate": 1.577766672219446e-05, - "loss": 0.1397, + "epoch": 0.42, + "learning_rate": 1.5776321634828965e-05, + "loss": 0.0874, "step": 3815 }, { - "epoch": 0.21, - "learning_rate": 1.5772113943028485e-05, - "loss": 0.2247, + "epoch": 0.42, + "learning_rate": 1.5770768547312308e-05, + "loss": 0.0904, "step": 3820 }, { - "epoch": 0.21, - "learning_rate": 1.5766561163862514e-05, - "loss": 0.1239, + "epoch": 0.42, + "learning_rate": 1.5765215459795647e-05, + "loss": 0.0897, "step": 3825 }, { - "epoch": 0.21, - "learning_rate": 1.5761008384696543e-05, - "loss": 0.1023, + "epoch": 0.43, + "learning_rate": 1.5759662372278987e-05, + "loss": 0.0367, "step": 3830 }, { - "epoch": 0.21, - "learning_rate": 1.575545560553057e-05, - "loss": 0.0824, + "epoch": 0.43, + "learning_rate": 1.575410928476233e-05, + "loss": 0.0713, "step": 3835 }, { - "epoch": 0.21, - "learning_rate": 1.5749902826364596e-05, - "loss": 0.1347, + "epoch": 0.43, + "learning_rate": 1.574855619724567e-05, + "loss": 0.1449, "step": 3840 }, { - "epoch": 0.21, - "learning_rate": 1.5744350047198625e-05, - "loss": 0.221, + "epoch": 0.43, + "learning_rate": 1.574300310972901e-05, + "loss": 0.1427, "step": 3845 }, { - "epoch": 0.21, - "learning_rate": 1.573879726803265e-05, - "loss": 0.5271, + "epoch": 0.43, + "learning_rate": 1.573745002221235e-05, + "loss": 0.2691, "step": 3850 }, { - "epoch": 0.21, - "learning_rate": 1.573324448886668e-05, - "loss": 0.128, + "epoch": 0.43, + "learning_rate": 1.573189693469569e-05, + "loss": 0.1211, "step": 3855 }, { - "epoch": 0.21, - "learning_rate": 1.5727691709700708e-05, - "loss": 0.1069, + "epoch": 0.43, + "learning_rate": 1.5726343847179033e-05, + "loss": 0.0944, "step": 3860 }, { - "epoch": 0.21, - "learning_rate": 1.5722138930534733e-05, - "loss": 0.1632, + "epoch": 0.43, + "learning_rate": 1.5720790759662376e-05, + "loss": 0.1037, "step": 3865 }, { - "epoch": 0.21, - "learning_rate": 1.571658615136876e-05, - "loss": 0.1169, + "epoch": 0.43, + "learning_rate": 1.5715237672145712e-05, + "loss": 0.08, "step": 3870 }, { - "epoch": 0.22, - "learning_rate": 1.571103337220279e-05, - "loss": 0.1632, + "epoch": 0.43, + "learning_rate": 1.5709684584629055e-05, + "loss": 0.1048, "step": 3875 }, { - "epoch": 0.22, - "learning_rate": 1.5705480593036815e-05, - "loss": 0.058, + "epoch": 0.43, + "learning_rate": 1.5704131497112397e-05, + "loss": 0.0437, "step": 3880 }, { - "epoch": 0.22, - "learning_rate": 1.5699927813870844e-05, - "loss": 0.1595, + "epoch": 0.43, + "learning_rate": 1.5698578409595737e-05, + "loss": 0.0731, "step": 3885 }, { - "epoch": 0.22, - "learning_rate": 1.5694375034704873e-05, - "loss": 0.1721, + "epoch": 0.43, + "learning_rate": 1.5693025322079076e-05, + "loss": 0.1378, "step": 3890 }, { - "epoch": 0.22, - "learning_rate": 1.5688822255538898e-05, - "loss": 0.159, + "epoch": 0.43, + "learning_rate": 1.568747223456242e-05, + "loss": 0.1637, "step": 3895 }, { - "epoch": 0.22, - "learning_rate": 1.5683269476372927e-05, - "loss": 0.3106, + "epoch": 0.43, + "learning_rate": 1.5681919147045758e-05, + "loss": 0.2737, "step": 3900 }, { - "epoch": 0.22, - "learning_rate": 1.5677716697206952e-05, - "loss": 0.1332, + "epoch": 0.43, + "learning_rate": 1.56763660595291e-05, + "loss": 0.0846, "step": 3905 }, { - "epoch": 0.22, - "learning_rate": 1.567216391804098e-05, - "loss": 0.1309, + "epoch": 0.43, + "learning_rate": 1.567081297201244e-05, + "loss": 0.0757, "step": 3910 }, { - "epoch": 0.22, - "learning_rate": 1.566661113887501e-05, - "loss": 0.1314, + "epoch": 0.43, + "learning_rate": 1.566525988449578e-05, + "loss": 0.09, "step": 3915 }, { - "epoch": 0.22, - "learning_rate": 1.5661058359709034e-05, - "loss": 0.1286, + "epoch": 0.44, + "learning_rate": 1.5659706796979122e-05, + "loss": 0.1917, "step": 3920 }, { - "epoch": 0.22, - "learning_rate": 1.5655505580543063e-05, - "loss": 0.125, + "epoch": 0.44, + "learning_rate": 1.5654153709462462e-05, + "loss": 0.0921, "step": 3925 }, { - "epoch": 0.22, - "learning_rate": 1.5649952801377092e-05, - "loss": 0.0657, + "epoch": 0.44, + "learning_rate": 1.5648600621945805e-05, + "loss": 0.0803, "step": 3930 }, { - "epoch": 0.22, - "learning_rate": 1.5644400022211117e-05, - "loss": 0.272, + "epoch": 0.44, + "learning_rate": 1.5643047534429144e-05, + "loss": 0.1142, "step": 3935 }, { - "epoch": 0.22, - "learning_rate": 1.5638847243045146e-05, - "loss": 0.2074, + "epoch": 0.44, + "learning_rate": 1.5637494446912483e-05, + "loss": 0.1233, "step": 3940 }, { - "epoch": 0.22, - "learning_rate": 1.563329446387917e-05, - "loss": 0.2605, + "epoch": 0.44, + "learning_rate": 1.5631941359395826e-05, + "loss": 0.1779, "step": 3945 }, { - "epoch": 0.22, - "learning_rate": 1.56277416847132e-05, - "loss": 0.4193, + "epoch": 0.44, + "learning_rate": 1.5626388271879165e-05, + "loss": 0.305, "step": 3950 }, { - "epoch": 0.22, - "learning_rate": 1.5622188905547228e-05, - "loss": 0.171, + "epoch": 0.44, + "learning_rate": 1.5620835184362505e-05, + "loss": 0.1212, "step": 3955 }, { - "epoch": 0.22, - "learning_rate": 1.5616636126381253e-05, - "loss": 0.1672, + "epoch": 0.44, + "learning_rate": 1.5615282096845848e-05, + "loss": 0.0716, "step": 3960 }, { - "epoch": 0.22, - "learning_rate": 1.5611083347215282e-05, - "loss": 0.1428, + "epoch": 0.44, + "learning_rate": 1.5609729009329187e-05, + "loss": 0.0825, "step": 3965 }, { - "epoch": 0.22, - "learning_rate": 1.560553056804931e-05, - "loss": 0.166, + "epoch": 0.44, + "learning_rate": 1.560417592181253e-05, + "loss": 0.13, "step": 3970 }, { - "epoch": 0.22, - "learning_rate": 1.559997778888334e-05, - "loss": 0.0772, + "epoch": 0.44, + "learning_rate": 1.559862283429587e-05, + "loss": 0.0773, "step": 3975 }, { - "epoch": 0.22, - "learning_rate": 1.5594425009717365e-05, - "loss": 0.0507, + "epoch": 0.44, + "learning_rate": 1.5593069746779212e-05, + "loss": 0.0568, "step": 3980 }, { - "epoch": 0.22, - "learning_rate": 1.5588872230551393e-05, - "loss": 0.1049, + "epoch": 0.44, + "learning_rate": 1.558751665926255e-05, + "loss": 0.0973, "step": 3985 }, { - "epoch": 0.22, - "learning_rate": 1.558331945138542e-05, - "loss": 0.1257, + "epoch": 0.44, + "learning_rate": 1.5581963571745894e-05, + "loss": 0.1212, "step": 3990 }, { - "epoch": 0.22, - "learning_rate": 1.5577766672219447e-05, - "loss": 0.2309, + "epoch": 0.44, + "learning_rate": 1.5576410484229233e-05, + "loss": 0.2039, "step": 3995 }, { - "epoch": 0.22, - "learning_rate": 1.5572213893053472e-05, - "loss": 0.4747, + "epoch": 0.44, + "learning_rate": 1.5570857396712573e-05, + "loss": 0.2844, "step": 4000 }, { - "epoch": 0.22, - "eval_loss": 0.07289480417966843, - "eval_runtime": 1725.6696, - "eval_samples_per_second": 9.282, - "eval_steps_per_second": 2.321, - "eval_wer": 0.20290299314821492, + "epoch": 0.44, + "eval_loss": 0.039118871092796326, + "eval_runtime": 1816.348, + "eval_samples_per_second": 8.819, + "eval_steps_per_second": 2.205, + "eval_wer": 0.10885322755138839, "step": 4000 }, { - "epoch": 0.22, - "learning_rate": 1.55666611138875e-05, - "loss": 0.1549, + "epoch": 0.44, + "learning_rate": 1.5565304309195915e-05, + "loss": 0.0808, "step": 4005 }, { - "epoch": 0.22, - "learning_rate": 1.556110833472153e-05, - "loss": 0.1362, + "epoch": 0.45, + "learning_rate": 1.5559751221679255e-05, + "loss": 0.0893, "step": 4010 }, { - "epoch": 0.22, - "learning_rate": 1.555555555555556e-05, - "loss": 0.1429, + "epoch": 0.45, + "learning_rate": 1.5554198134162594e-05, + "loss": 0.1097, "step": 4015 }, { - "epoch": 0.22, - "learning_rate": 1.5550002776389584e-05, - "loss": 0.1351, + "epoch": 0.45, + "learning_rate": 1.5548645046645937e-05, + "loss": 0.1095, "step": 4020 }, { - "epoch": 0.22, - "learning_rate": 1.5544449997223612e-05, - "loss": 0.1408, + "epoch": 0.45, + "learning_rate": 1.5543091959129276e-05, + "loss": 0.0481, "step": 4025 }, { - "epoch": 0.22, - "learning_rate": 1.5538897218057637e-05, - "loss": 0.0407, + "epoch": 0.45, + "learning_rate": 1.553753887161262e-05, + "loss": 0.053, "step": 4030 }, { - "epoch": 0.22, - "learning_rate": 1.5533344438891666e-05, - "loss": 0.1208, + "epoch": 0.45, + "learning_rate": 1.553198578409596e-05, + "loss": 0.0695, "step": 4035 }, { - "epoch": 0.22, - "learning_rate": 1.5527791659725695e-05, - "loss": 0.1584, + "epoch": 0.45, + "learning_rate": 1.5526432696579298e-05, + "loss": 0.0851, "step": 4040 }, { - "epoch": 0.22, - "learning_rate": 1.552223888055972e-05, - "loss": 0.1945, + "epoch": 0.45, + "learning_rate": 1.552087960906264e-05, + "loss": 0.2526, "step": 4045 }, { - "epoch": 0.22, - "learning_rate": 1.551668610139375e-05, - "loss": 0.4511, + "epoch": 0.45, + "learning_rate": 1.551532652154598e-05, + "loss": 0.2974, "step": 4050 }, { - "epoch": 0.23, - "learning_rate": 1.5511133322227777e-05, - "loss": 0.1241, + "epoch": 0.45, + "learning_rate": 1.5509773434029323e-05, + "loss": 0.0649, "step": 4055 }, { - "epoch": 0.23, - "learning_rate": 1.5505580543061806e-05, - "loss": 0.1402, + "epoch": 0.45, + "learning_rate": 1.5504220346512662e-05, + "loss": 0.083, "step": 4060 }, { - "epoch": 0.23, - "learning_rate": 1.550002776389583e-05, - "loss": 0.1457, + "epoch": 0.45, + "learning_rate": 1.5498667258996e-05, + "loss": 0.1154, "step": 4065 }, { - "epoch": 0.23, - "learning_rate": 1.549447498472986e-05, - "loss": 0.1405, + "epoch": 0.45, + "learning_rate": 1.5493114171479344e-05, + "loss": 0.1205, "step": 4070 }, { - "epoch": 0.23, - "learning_rate": 1.5488922205563885e-05, - "loss": 0.1723, + "epoch": 0.45, + "learning_rate": 1.5487561083962687e-05, + "loss": 0.0725, "step": 4075 }, { - "epoch": 0.23, - "learning_rate": 1.5483369426397914e-05, - "loss": 0.0592, + "epoch": 0.45, + "learning_rate": 1.5482007996446023e-05, + "loss": 0.0479, "step": 4080 }, { - "epoch": 0.23, - "learning_rate": 1.547781664723194e-05, - "loss": 0.0893, + "epoch": 0.45, + "learning_rate": 1.5476454908929366e-05, + "loss": 0.0567, "step": 4085 }, { - "epoch": 0.23, - "learning_rate": 1.5472263868065968e-05, - "loss": 0.1843, + "epoch": 0.45, + "learning_rate": 1.547090182141271e-05, + "loss": 0.0601, "step": 4090 }, { - "epoch": 0.23, - "learning_rate": 1.5466711088899996e-05, - "loss": 0.2373, + "epoch": 0.45, + "learning_rate": 1.5465348733896048e-05, + "loss": 0.1391, "step": 4095 }, { - "epoch": 0.23, - "learning_rate": 1.5461158309734025e-05, - "loss": 0.376, + "epoch": 0.46, + "learning_rate": 1.5459795646379387e-05, + "loss": 0.2792, "step": 4100 }, { - "epoch": 0.23, - "learning_rate": 1.545560553056805e-05, - "loss": 0.0867, + "epoch": 0.46, + "learning_rate": 1.545424255886273e-05, + "loss": 0.0962, "step": 4105 }, { - "epoch": 0.23, - "learning_rate": 1.545005275140208e-05, - "loss": 0.1096, + "epoch": 0.46, + "learning_rate": 1.544868947134607e-05, + "loss": 0.095, "step": 4110 }, { - "epoch": 0.23, - "learning_rate": 1.5444499972236104e-05, - "loss": 0.1957, + "epoch": 0.46, + "learning_rate": 1.5443136383829412e-05, + "loss": 0.093, "step": 4115 }, { - "epoch": 0.23, - "learning_rate": 1.5438947193070133e-05, - "loss": 0.1902, + "epoch": 0.46, + "learning_rate": 1.543758329631275e-05, + "loss": 0.0956, "step": 4120 }, { - "epoch": 0.23, - "learning_rate": 1.543339441390416e-05, - "loss": 0.0836, + "epoch": 0.46, + "learning_rate": 1.543203020879609e-05, + "loss": 0.0835, "step": 4125 }, { - "epoch": 0.23, - "learning_rate": 1.5427841634738187e-05, - "loss": 0.0962, + "epoch": 0.46, + "learning_rate": 1.5426477121279434e-05, + "loss": 0.0798, "step": 4130 }, { - "epoch": 0.23, - "learning_rate": 1.5422288855572215e-05, - "loss": 0.1164, + "epoch": 0.46, + "learning_rate": 1.5420924033762773e-05, + "loss": 0.1073, "step": 4135 }, { - "epoch": 0.23, - "learning_rate": 1.5416736076406244e-05, - "loss": 0.1613, + "epoch": 0.46, + "learning_rate": 1.5415370946246112e-05, + "loss": 0.106, "step": 4140 }, { - "epoch": 0.23, - "learning_rate": 1.541118329724027e-05, - "loss": 0.2705, + "epoch": 0.46, + "learning_rate": 1.5409817858729455e-05, + "loss": 0.2011, "step": 4145 }, { - "epoch": 0.23, - "learning_rate": 1.5405630518074298e-05, - "loss": 0.6455, + "epoch": 0.46, + "learning_rate": 1.5404264771212794e-05, + "loss": 0.4367, "step": 4150 }, { - "epoch": 0.23, - "learning_rate": 1.5400077738908326e-05, - "loss": 0.1278, + "epoch": 0.46, + "learning_rate": 1.5398711683696137e-05, + "loss": 0.0719, "step": 4155 }, { - "epoch": 0.23, - "learning_rate": 1.5394524959742352e-05, - "loss": 0.1229, + "epoch": 0.46, + "learning_rate": 1.5393158596179477e-05, + "loss": 0.0725, "step": 4160 }, { - "epoch": 0.23, - "learning_rate": 1.538897218057638e-05, - "loss": 0.1649, + "epoch": 0.46, + "learning_rate": 1.5387605508662816e-05, + "loss": 0.1205, "step": 4165 }, { - "epoch": 0.23, - "learning_rate": 1.5383419401410406e-05, - "loss": 0.1492, + "epoch": 0.46, + "learning_rate": 1.538205242114616e-05, + "loss": 0.0822, "step": 4170 }, { - "epoch": 0.23, - "learning_rate": 1.5377866622244434e-05, - "loss": 0.1179, + "epoch": 0.46, + "learning_rate": 1.53764993336295e-05, + "loss": 0.0884, "step": 4175 }, { - "epoch": 0.23, - "learning_rate": 1.5372313843078463e-05, - "loss": 0.1114, + "epoch": 0.46, + "learning_rate": 1.5370946246112837e-05, + "loss": 0.0544, "step": 4180 }, { - "epoch": 0.23, - "learning_rate": 1.5366761063912488e-05, - "loss": 0.0854, + "epoch": 0.46, + "learning_rate": 1.536539315859618e-05, + "loss": 0.0357, "step": 4185 }, { - "epoch": 0.23, - "learning_rate": 1.5361208284746517e-05, - "loss": 0.1621, + "epoch": 0.47, + "learning_rate": 1.5359840071079523e-05, + "loss": 0.1223, "step": 4190 }, { - "epoch": 0.23, - "learning_rate": 1.5355655505580545e-05, - "loss": 0.1695, + "epoch": 0.47, + "learning_rate": 1.5354286983562862e-05, + "loss": 0.156, "step": 4195 }, { - "epoch": 0.23, - "learning_rate": 1.535010272641457e-05, - "loss": 0.5718, + "epoch": 0.47, + "learning_rate": 1.5348733896046205e-05, + "loss": 0.3048, "step": 4200 }, { - "epoch": 0.23, - "learning_rate": 1.53445499472486e-05, - "loss": 0.1023, + "epoch": 0.47, + "learning_rate": 1.5343180808529544e-05, + "loss": 0.0934, "step": 4205 }, { - "epoch": 0.23, - "learning_rate": 1.5338997168082625e-05, - "loss": 0.1503, + "epoch": 0.47, + "learning_rate": 1.5337627721012884e-05, + "loss": 0.1207, "step": 4210 }, { - "epoch": 0.23, - "learning_rate": 1.5333444388916653e-05, - "loss": 0.1333, + "epoch": 0.47, + "learning_rate": 1.5332074633496227e-05, + "loss": 0.0921, "step": 4215 }, { - "epoch": 0.23, - "learning_rate": 1.5327891609750682e-05, - "loss": 0.1594, + "epoch": 0.47, + "learning_rate": 1.5326521545979566e-05, + "loss": 0.1101, "step": 4220 }, { - "epoch": 0.23, - "learning_rate": 1.5322338830584707e-05, - "loss": 0.1238, + "epoch": 0.47, + "learning_rate": 1.5320968458462905e-05, + "loss": 0.0986, "step": 4225 }, { - "epoch": 0.23, - "learning_rate": 1.5316786051418736e-05, - "loss": 0.1006, + "epoch": 0.47, + "learning_rate": 1.5315415370946248e-05, + "loss": 0.0326, "step": 4230 }, { - "epoch": 0.24, - "learning_rate": 1.5311233272252764e-05, - "loss": 0.1289, + "epoch": 0.47, + "learning_rate": 1.5309862283429587e-05, + "loss": 0.0517, "step": 4235 }, { - "epoch": 0.24, - "learning_rate": 1.5305680493086793e-05, - "loss": 0.1162, + "epoch": 0.47, + "learning_rate": 1.530430919591293e-05, + "loss": 0.0859, "step": 4240 }, { - "epoch": 0.24, - "learning_rate": 1.5300127713920818e-05, - "loss": 0.1831, + "epoch": 0.47, + "learning_rate": 1.529875610839627e-05, + "loss": 0.1802, "step": 4245 }, { - "epoch": 0.24, - "learning_rate": 1.5294574934754847e-05, - "loss": 0.3605, + "epoch": 0.47, + "learning_rate": 1.529320302087961e-05, + "loss": 0.3513, "step": 4250 }, { - "epoch": 0.24, - "learning_rate": 1.5289022155588872e-05, - "loss": 0.1715, + "epoch": 0.47, + "learning_rate": 1.528764993336295e-05, + "loss": 0.0682, "step": 4255 }, { - "epoch": 0.24, - "learning_rate": 1.52834693764229e-05, - "loss": 0.1148, + "epoch": 0.47, + "learning_rate": 1.528209684584629e-05, + "loss": 0.1053, "step": 4260 }, { - "epoch": 0.24, - "learning_rate": 1.5277916597256926e-05, - "loss": 0.1363, + "epoch": 0.47, + "learning_rate": 1.527654375832963e-05, + "loss": 0.1228, "step": 4265 }, { - "epoch": 0.24, - "learning_rate": 1.5272363818090955e-05, - "loss": 0.1247, + "epoch": 0.47, + "learning_rate": 1.5270990670812973e-05, + "loss": 0.1026, "step": 4270 }, { - "epoch": 0.24, - "learning_rate": 1.5266811038924983e-05, - "loss": 0.1438, + "epoch": 0.47, + "learning_rate": 1.5265437583296313e-05, + "loss": 0.0545, "step": 4275 }, { - "epoch": 0.24, - "learning_rate": 1.5261258259759012e-05, - "loss": 0.1258, + "epoch": 0.48, + "learning_rate": 1.5259884495779655e-05, + "loss": 0.0536, "step": 4280 }, { - "epoch": 0.24, - "learning_rate": 1.5255705480593037e-05, - "loss": 0.2283, + "epoch": 0.48, + "learning_rate": 1.5254331408262995e-05, + "loss": 0.0977, "step": 4285 }, { - "epoch": 0.24, - "learning_rate": 1.5250152701427066e-05, - "loss": 0.1031, + "epoch": 0.48, + "learning_rate": 1.5248778320746336e-05, + "loss": 0.1605, "step": 4290 }, { - "epoch": 0.24, - "learning_rate": 1.5244599922261091e-05, - "loss": 0.2167, + "epoch": 0.48, + "learning_rate": 1.5243225233229677e-05, + "loss": 0.1559, "step": 4295 }, { - "epoch": 0.24, - "learning_rate": 1.523904714309512e-05, - "loss": 0.4091, + "epoch": 0.48, + "learning_rate": 1.5237672145713018e-05, + "loss": 0.3257, "step": 4300 }, { - "epoch": 0.24, - "learning_rate": 1.5233494363929148e-05, - "loss": 0.1286, + "epoch": 0.48, + "learning_rate": 1.5232119058196357e-05, + "loss": 0.0931, "step": 4305 }, { - "epoch": 0.24, - "learning_rate": 1.5227941584763175e-05, - "loss": 0.1233, + "epoch": 0.48, + "learning_rate": 1.5226565970679698e-05, + "loss": 0.0717, "step": 4310 }, { - "epoch": 0.24, - "learning_rate": 1.5222388805597204e-05, - "loss": 0.1337, + "epoch": 0.48, + "learning_rate": 1.522101288316304e-05, + "loss": 0.0744, "step": 4315 }, { - "epoch": 0.24, - "learning_rate": 1.521683602643123e-05, - "loss": 0.1575, + "epoch": 0.48, + "learning_rate": 1.521545979564638e-05, + "loss": 0.1406, "step": 4320 }, { - "epoch": 0.24, - "learning_rate": 1.5211283247265258e-05, - "loss": 0.1059, + "epoch": 0.48, + "learning_rate": 1.5209906708129723e-05, + "loss": 0.0879, "step": 4325 }, { - "epoch": 0.24, - "learning_rate": 1.5205730468099285e-05, - "loss": 0.0598, + "epoch": 0.48, + "learning_rate": 1.520435362061306e-05, + "loss": 0.0413, "step": 4330 }, { - "epoch": 0.24, - "learning_rate": 1.5200177688933314e-05, - "loss": 0.2193, + "epoch": 0.48, + "learning_rate": 1.5198800533096404e-05, + "loss": 0.0539, "step": 4335 }, { - "epoch": 0.24, - "learning_rate": 1.5194624909767339e-05, - "loss": 0.2097, + "epoch": 0.48, + "learning_rate": 1.5193247445579745e-05, + "loss": 0.087, "step": 4340 }, { - "epoch": 0.24, - "learning_rate": 1.5189072130601367e-05, - "loss": 0.2476, + "epoch": 0.48, + "learning_rate": 1.5187694358063086e-05, + "loss": 0.1241, "step": 4345 }, { - "epoch": 0.24, - "learning_rate": 1.5183519351435394e-05, - "loss": 0.364, + "epoch": 0.48, + "learning_rate": 1.5182141270546425e-05, + "loss": 0.2865, "step": 4350 }, { - "epoch": 0.24, - "learning_rate": 1.5177966572269423e-05, - "loss": 0.1281, + "epoch": 0.48, + "learning_rate": 1.5176588183029766e-05, + "loss": 0.0835, "step": 4355 }, { - "epoch": 0.24, - "learning_rate": 1.5172413793103448e-05, - "loss": 0.1348, + "epoch": 0.48, + "learning_rate": 1.5171035095513107e-05, + "loss": 0.0901, "step": 4360 }, { - "epoch": 0.24, - "learning_rate": 1.5166861013937477e-05, - "loss": 0.1599, + "epoch": 0.48, + "learning_rate": 1.5165482007996448e-05, + "loss": 0.1145, "step": 4365 }, { - "epoch": 0.24, - "learning_rate": 1.5161308234771504e-05, - "loss": 0.1901, + "epoch": 0.49, + "learning_rate": 1.5159928920479788e-05, + "loss": 0.1068, "step": 4370 }, { - "epoch": 0.24, - "learning_rate": 1.5155755455605533e-05, - "loss": 0.1344, + "epoch": 0.49, + "learning_rate": 1.5154375832963129e-05, + "loss": 0.0441, "step": 4375 }, { - "epoch": 0.24, - "learning_rate": 1.5150202676439558e-05, - "loss": 0.0926, + "epoch": 0.49, + "learning_rate": 1.514882274544647e-05, + "loss": 0.0632, "step": 4380 }, { - "epoch": 0.24, - "learning_rate": 1.5144649897273586e-05, - "loss": 0.0776, + "epoch": 0.49, + "learning_rate": 1.514326965792981e-05, + "loss": 0.1224, "step": 4385 }, { - "epoch": 0.24, - "learning_rate": 1.5139097118107615e-05, - "loss": 0.2744, + "epoch": 0.49, + "learning_rate": 1.513771657041315e-05, + "loss": 0.1325, "step": 4390 }, { - "epoch": 0.24, - "learning_rate": 1.5133544338941642e-05, - "loss": 0.1109, + "epoch": 0.49, + "learning_rate": 1.5132163482896491e-05, + "loss": 0.1927, "step": 4395 }, { - "epoch": 0.24, - "learning_rate": 1.512799155977567e-05, - "loss": 0.4835, + "epoch": 0.49, + "learning_rate": 1.5126610395379832e-05, + "loss": 0.2885, "step": 4400 }, { - "epoch": 0.24, - "learning_rate": 1.5122438780609696e-05, - "loss": 0.1005, + "epoch": 0.49, + "learning_rate": 1.5121057307863173e-05, + "loss": 0.0798, "step": 4405 }, { - "epoch": 0.24, - "learning_rate": 1.5116886001443725e-05, - "loss": 0.1367, + "epoch": 0.49, + "learning_rate": 1.5115504220346513e-05, + "loss": 0.087, "step": 4410 }, { - "epoch": 0.25, - "learning_rate": 1.5111333222277752e-05, - "loss": 0.1679, + "epoch": 0.49, + "learning_rate": 1.5109951132829854e-05, + "loss": 0.0931, "step": 4415 }, { - "epoch": 0.25, - "learning_rate": 1.510578044311178e-05, - "loss": 0.1314, + "epoch": 0.49, + "learning_rate": 1.5104398045313195e-05, + "loss": 0.0984, "step": 4420 }, { - "epoch": 0.25, - "learning_rate": 1.5100227663945805e-05, - "loss": 0.1004, + "epoch": 0.49, + "learning_rate": 1.5098844957796538e-05, + "loss": 0.0624, "step": 4425 }, { - "epoch": 0.25, - "learning_rate": 1.5094674884779834e-05, - "loss": 0.0736, + "epoch": 0.49, + "learning_rate": 1.5093291870279875e-05, + "loss": 0.0547, "step": 4430 }, { - "epoch": 0.25, - "learning_rate": 1.5089122105613861e-05, - "loss": 0.1012, + "epoch": 0.49, + "learning_rate": 1.5087738782763216e-05, + "loss": 0.094, "step": 4435 }, { - "epoch": 0.25, - "learning_rate": 1.508356932644789e-05, - "loss": 0.1648, + "epoch": 0.49, + "learning_rate": 1.5082185695246559e-05, + "loss": 0.0959, "step": 4440 }, { - "epoch": 0.25, - "learning_rate": 1.5078016547281915e-05, - "loss": 0.2751, + "epoch": 0.49, + "learning_rate": 1.50766326077299e-05, + "loss": 0.1507, "step": 4445 }, { - "epoch": 0.25, - "learning_rate": 1.5072463768115944e-05, - "loss": 0.4631, + "epoch": 0.49, + "learning_rate": 1.507107952021324e-05, + "loss": 0.2696, "step": 4450 }, { - "epoch": 0.25, - "learning_rate": 1.506691098894997e-05, - "loss": 0.1518, + "epoch": 0.49, + "learning_rate": 1.506552643269658e-05, + "loss": 0.0944, "step": 4455 }, { - "epoch": 0.25, - "learning_rate": 1.5061358209783999e-05, - "loss": 0.1369, + "epoch": 0.5, + "learning_rate": 1.5059973345179922e-05, + "loss": 0.0857, "step": 4460 }, { - "epoch": 0.25, - "learning_rate": 1.5055805430618024e-05, - "loss": 0.144, + "epoch": 0.5, + "learning_rate": 1.5054420257663263e-05, + "loss": 0.0834, "step": 4465 }, { - "epoch": 0.25, - "learning_rate": 1.5050252651452053e-05, - "loss": 0.1767, + "epoch": 0.5, + "learning_rate": 1.5048867170146604e-05, + "loss": 0.1062, "step": 4470 }, { - "epoch": 0.25, - "learning_rate": 1.504469987228608e-05, - "loss": 0.104, + "epoch": 0.5, + "learning_rate": 1.5043314082629943e-05, + "loss": 0.0683, "step": 4475 }, { - "epoch": 0.25, - "learning_rate": 1.5039147093120109e-05, - "loss": 0.0653, + "epoch": 0.5, + "learning_rate": 1.5037760995113284e-05, + "loss": 0.0512, "step": 4480 }, { - "epoch": 0.25, - "learning_rate": 1.5033594313954136e-05, - "loss": 0.0804, + "epoch": 0.5, + "learning_rate": 1.5032207907596625e-05, + "loss": 0.0896, "step": 4485 }, { - "epoch": 0.25, - "learning_rate": 1.5028041534788163e-05, - "loss": 0.1759, + "epoch": 0.5, + "learning_rate": 1.5026654820079966e-05, + "loss": 0.1176, "step": 4490 }, { - "epoch": 0.25, - "learning_rate": 1.5022488755622191e-05, - "loss": 0.2305, + "epoch": 0.5, + "learning_rate": 1.5022212350066638e-05, + "loss": 0.3841, "step": 4495 }, { - "epoch": 0.25, - "learning_rate": 1.5016935976456218e-05, - "loss": 0.3476, + "epoch": 0.5, + "learning_rate": 1.5016659262549979e-05, + "loss": 0.3718, "step": 4500 }, { - "epoch": 0.25, - "learning_rate": 1.5011383197290245e-05, - "loss": 0.092, + "epoch": 0.5, + "learning_rate": 1.5011106175033318e-05, + "loss": 0.0946, "step": 4505 }, { - "epoch": 0.25, - "learning_rate": 1.5005830418124272e-05, - "loss": 0.1192, + "epoch": 0.5, + "learning_rate": 1.500555308751666e-05, + "loss": 0.058, "step": 4510 }, { - "epoch": 0.25, - "learning_rate": 1.50002776389583e-05, - "loss": 0.122, + "epoch": 0.5, + "learning_rate": 1.5000000000000002e-05, + "loss": 0.1082, "step": 4515 }, { - "epoch": 0.25, - "learning_rate": 1.4994724859792326e-05, - "loss": 0.1426, + "epoch": 0.5, + "learning_rate": 1.4994446912483343e-05, + "loss": 0.1132, "step": 4520 }, { - "epoch": 0.25, - "learning_rate": 1.4989172080626355e-05, - "loss": 0.1157, + "epoch": 0.5, + "learning_rate": 1.4988893824966684e-05, + "loss": 0.0723, "step": 4525 }, { - "epoch": 0.25, - "learning_rate": 1.4983619301460381e-05, - "loss": 0.0837, + "epoch": 0.5, + "learning_rate": 1.4983340737450024e-05, + "loss": 0.0698, "step": 4530 }, { - "epoch": 0.25, - "learning_rate": 1.497806652229441e-05, - "loss": 0.067, + "epoch": 0.5, + "learning_rate": 1.4977787649933365e-05, + "loss": 0.0791, "step": 4535 }, { - "epoch": 0.25, - "learning_rate": 1.4972513743128435e-05, - "loss": 0.0884, + "epoch": 0.5, + "learning_rate": 1.4972234562416706e-05, + "loss": 0.0893, "step": 4540 }, { - "epoch": 0.25, - "learning_rate": 1.4966960963962464e-05, - "loss": 0.3156, + "epoch": 0.5, + "learning_rate": 1.4966681474900047e-05, + "loss": 0.1476, "step": 4545 }, { - "epoch": 0.25, - "learning_rate": 1.4961408184796491e-05, - "loss": 0.5914, + "epoch": 0.51, + "learning_rate": 1.4961128387383386e-05, + "loss": 0.3097, "step": 4550 }, { - "epoch": 0.25, - "learning_rate": 1.495585540563052e-05, - "loss": 0.1523, + "epoch": 0.51, + "learning_rate": 1.4955575299866727e-05, + "loss": 0.0913, "step": 4555 }, { - "epoch": 0.25, - "learning_rate": 1.4950302626464545e-05, - "loss": 0.1353, + "epoch": 0.51, + "learning_rate": 1.4950022212350068e-05, + "loss": 0.09, "step": 4560 }, { - "epoch": 0.25, - "learning_rate": 1.4944749847298574e-05, - "loss": 0.1458, + "epoch": 0.51, + "learning_rate": 1.494446912483341e-05, + "loss": 0.1254, "step": 4565 }, { - "epoch": 0.25, - "learning_rate": 1.4939197068132602e-05, - "loss": 0.1743, + "epoch": 0.51, + "learning_rate": 1.4938916037316749e-05, + "loss": 0.0895, "step": 4570 }, { - "epoch": 0.25, - "learning_rate": 1.4933644288966629e-05, - "loss": 0.1095, + "epoch": 0.51, + "learning_rate": 1.493336294980009e-05, + "loss": 0.0573, "step": 4575 }, { - "epoch": 0.25, - "learning_rate": 1.4928091509800658e-05, - "loss": 0.0651, + "epoch": 0.51, + "learning_rate": 1.492780986228343e-05, + "loss": 0.049, "step": 4580 }, { - "epoch": 0.25, - "learning_rate": 1.4922538730634683e-05, - "loss": 0.1243, + "epoch": 0.51, + "learning_rate": 1.4922256774766772e-05, + "loss": 0.0823, "step": 4585 }, { - "epoch": 0.25, - "learning_rate": 1.4916985951468712e-05, - "loss": 0.1176, + "epoch": 0.51, + "learning_rate": 1.4916703687250111e-05, + "loss": 0.123, "step": 4590 }, { - "epoch": 0.26, - "learning_rate": 1.4911433172302739e-05, - "loss": 0.2488, + "epoch": 0.51, + "learning_rate": 1.4911150599733452e-05, + "loss": 0.1552, "step": 4595 }, { - "epoch": 0.26, - "learning_rate": 1.4905880393136767e-05, - "loss": 0.3971, + "epoch": 0.51, + "learning_rate": 1.4905597512216793e-05, + "loss": 0.4451, "step": 4600 }, { - "epoch": 0.26, - "learning_rate": 1.4900327613970792e-05, - "loss": 0.1348, + "epoch": 0.51, + "learning_rate": 1.4900044424700134e-05, + "loss": 0.1121, "step": 4605 }, { - "epoch": 0.26, - "learning_rate": 1.4894774834804821e-05, - "loss": 0.1416, + "epoch": 0.51, + "learning_rate": 1.4894491337183474e-05, + "loss": 0.0816, "step": 4610 }, { - "epoch": 0.26, - "learning_rate": 1.4889222055638848e-05, - "loss": 0.1502, + "epoch": 0.51, + "learning_rate": 1.4888938249666815e-05, + "loss": 0.0919, "step": 4615 }, { - "epoch": 0.26, - "learning_rate": 1.4883669276472877e-05, - "loss": 0.1956, + "epoch": 0.51, + "learning_rate": 1.4883385162150158e-05, + "loss": 0.1133, "step": 4620 }, { - "epoch": 0.26, - "learning_rate": 1.4878116497306902e-05, - "loss": 0.0521, + "epoch": 0.51, + "learning_rate": 1.4877832074633499e-05, + "loss": 0.0803, "step": 4625 }, { - "epoch": 0.26, - "learning_rate": 1.487256371814093e-05, - "loss": 0.128, + "epoch": 0.51, + "learning_rate": 1.4872278987116836e-05, + "loss": 0.0438, "step": 4630 }, { - "epoch": 0.26, - "learning_rate": 1.4867010938974958e-05, - "loss": 0.1153, + "epoch": 0.51, + "learning_rate": 1.4866725899600179e-05, + "loss": 0.0981, "step": 4635 }, { - "epoch": 0.26, - "learning_rate": 1.4861458159808986e-05, - "loss": 0.1849, + "epoch": 0.51, + "learning_rate": 1.486117281208352e-05, + "loss": 0.0885, "step": 4640 }, { - "epoch": 0.26, - "learning_rate": 1.4855905380643011e-05, - "loss": 0.2209, + "epoch": 0.52, + "learning_rate": 1.4855619724566861e-05, + "loss": 0.2147, "step": 4645 }, { - "epoch": 0.26, - "learning_rate": 1.485035260147704e-05, - "loss": 0.2821, + "epoch": 0.52, + "learning_rate": 1.48500666370502e-05, + "loss": 0.2633, "step": 4650 }, { - "epoch": 0.26, - "learning_rate": 1.4844799822311069e-05, - "loss": 0.1056, + "epoch": 0.52, + "learning_rate": 1.4844513549533542e-05, + "loss": 0.0868, "step": 4655 }, { - "epoch": 0.26, - "learning_rate": 1.4839247043145096e-05, - "loss": 0.1116, + "epoch": 0.52, + "learning_rate": 1.4838960462016883e-05, + "loss": 0.0864, "step": 4660 }, { - "epoch": 0.26, - "learning_rate": 1.4833694263979124e-05, - "loss": 0.1235, + "epoch": 0.52, + "learning_rate": 1.4833407374500224e-05, + "loss": 0.0768, "step": 4665 }, { - "epoch": 0.26, - "learning_rate": 1.482814148481315e-05, - "loss": 0.1362, + "epoch": 0.52, + "learning_rate": 1.4827854286983565e-05, + "loss": 0.0965, "step": 4670 }, { - "epoch": 0.26, - "learning_rate": 1.4822588705647178e-05, - "loss": 0.1131, + "epoch": 0.52, + "learning_rate": 1.4822301199466904e-05, + "loss": 0.0779, "step": 4675 }, { - "epoch": 0.26, - "learning_rate": 1.4817035926481205e-05, - "loss": 0.0804, + "epoch": 0.52, + "learning_rate": 1.4816748111950245e-05, + "loss": 0.0463, "step": 4680 }, { - "epoch": 0.26, - "learning_rate": 1.4811483147315234e-05, - "loss": 0.1252, + "epoch": 0.52, + "learning_rate": 1.4811195024433586e-05, + "loss": 0.0687, "step": 4685 }, { - "epoch": 0.26, - "learning_rate": 1.4805930368149259e-05, - "loss": 0.1915, + "epoch": 0.52, + "learning_rate": 1.4805641936916927e-05, + "loss": 0.1446, "step": 4690 }, { - "epoch": 0.26, - "learning_rate": 1.4800377588983288e-05, - "loss": 0.2332, + "epoch": 0.52, + "learning_rate": 1.4800088849400267e-05, + "loss": 0.2321, "step": 4695 }, { - "epoch": 0.26, - "learning_rate": 1.4794824809817315e-05, - "loss": 0.369, + "epoch": 0.52, + "learning_rate": 1.4794535761883608e-05, + "loss": 0.2379, "step": 4700 }, { - "epoch": 0.26, - "learning_rate": 1.4789272030651343e-05, - "loss": 0.1165, + "epoch": 0.52, + "learning_rate": 1.4788982674366949e-05, + "loss": 0.0683, "step": 4705 }, { - "epoch": 0.26, - "learning_rate": 1.4783719251485369e-05, - "loss": 0.1123, + "epoch": 0.52, + "learning_rate": 1.4783429586850292e-05, + "loss": 0.0992, "step": 4710 }, { - "epoch": 0.26, - "learning_rate": 1.4778166472319397e-05, - "loss": 0.1536, + "epoch": 0.52, + "learning_rate": 1.477787649933363e-05, + "loss": 0.1148, "step": 4715 }, { - "epoch": 0.26, - "learning_rate": 1.4772613693153424e-05, - "loss": 0.1545, + "epoch": 0.52, + "learning_rate": 1.477232341181697e-05, + "loss": 0.0855, "step": 4720 }, { - "epoch": 0.26, - "learning_rate": 1.4767060913987453e-05, - "loss": 0.0453, + "epoch": 0.52, + "learning_rate": 1.4766770324300313e-05, + "loss": 0.0998, "step": 4725 }, { - "epoch": 0.26, - "learning_rate": 1.4761508134821478e-05, - "loss": 0.0842, + "epoch": 0.52, + "learning_rate": 1.4761217236783654e-05, + "loss": 0.042, "step": 4730 }, { - "epoch": 0.26, - "learning_rate": 1.4755955355655507e-05, - "loss": 0.1938, + "epoch": 0.53, + "learning_rate": 1.4755664149266994e-05, + "loss": 0.0834, "step": 4735 }, { - "epoch": 0.26, - "learning_rate": 1.4750402576489534e-05, - "loss": 0.2009, + "epoch": 0.53, + "learning_rate": 1.4750111061750335e-05, + "loss": 0.0889, "step": 4740 }, { - "epoch": 0.26, - "learning_rate": 1.4744849797323562e-05, - "loss": 0.1408, + "epoch": 0.53, + "learning_rate": 1.4744557974233676e-05, + "loss": 0.1453, "step": 4745 }, { - "epoch": 0.26, - "learning_rate": 1.473929701815759e-05, - "loss": 0.445, + "epoch": 0.53, + "learning_rate": 1.4739004886717017e-05, + "loss": 0.3652, "step": 4750 }, { - "epoch": 0.26, - "learning_rate": 1.4733744238991616e-05, - "loss": 0.1343, + "epoch": 0.53, + "learning_rate": 1.4733451799200356e-05, + "loss": 0.091, "step": 4755 }, { - "epoch": 0.26, - "learning_rate": 1.4728191459825645e-05, - "loss": 0.1143, + "epoch": 0.53, + "learning_rate": 1.4727898711683697e-05, + "loss": 0.0837, "step": 4760 }, { - "epoch": 0.26, - "learning_rate": 1.4722638680659672e-05, - "loss": 0.1479, + "epoch": 0.53, + "learning_rate": 1.4722345624167038e-05, + "loss": 0.0931, "step": 4765 }, { - "epoch": 0.26, - "learning_rate": 1.4717085901493699e-05, - "loss": 0.1494, + "epoch": 0.53, + "learning_rate": 1.471679253665038e-05, + "loss": 0.1036, "step": 4770 }, { - "epoch": 0.26, - "learning_rate": 1.4711533122327726e-05, - "loss": 0.0856, + "epoch": 0.53, + "learning_rate": 1.4711239449133719e-05, + "loss": 0.0945, "step": 4775 }, { - "epoch": 0.27, - "learning_rate": 1.4705980343161754e-05, - "loss": 0.1515, + "epoch": 0.53, + "learning_rate": 1.470568636161706e-05, + "loss": 0.0556, "step": 4780 }, { - "epoch": 0.27, - "learning_rate": 1.4700427563995781e-05, - "loss": 0.1707, + "epoch": 0.53, + "learning_rate": 1.47001332741004e-05, + "loss": 0.1127, "step": 4785 }, { - "epoch": 0.27, - "learning_rate": 1.4694874784829808e-05, - "loss": 0.1208, + "epoch": 0.53, + "learning_rate": 1.4694580186583742e-05, + "loss": 0.11, "step": 4790 }, { - "epoch": 0.27, - "learning_rate": 1.4689322005663835e-05, - "loss": 0.1512, + "epoch": 0.53, + "learning_rate": 1.4689027099067083e-05, + "loss": 0.1356, "step": 4795 }, { - "epoch": 0.27, - "learning_rate": 1.4683769226497864e-05, - "loss": 0.4179, + "epoch": 0.53, + "learning_rate": 1.4683474011550422e-05, + "loss": 0.3402, "step": 4800 }, { - "epoch": 0.27, - "learning_rate": 1.467821644733189e-05, - "loss": 0.1294, + "epoch": 0.53, + "learning_rate": 1.4677920924033763e-05, + "loss": 0.0831, "step": 4805 }, { - "epoch": 0.27, - "learning_rate": 1.4672663668165918e-05, - "loss": 0.1331, + "epoch": 0.53, + "learning_rate": 1.4672367836517104e-05, + "loss": 0.0801, "step": 4810 }, { - "epoch": 0.27, - "learning_rate": 1.4667110888999945e-05, - "loss": 0.1086, + "epoch": 0.53, + "learning_rate": 1.4666814749000447e-05, + "loss": 0.0765, "step": 4815 }, { - "epoch": 0.27, - "learning_rate": 1.4661558109833973e-05, - "loss": 0.1096, + "epoch": 0.53, + "learning_rate": 1.4661261661483785e-05, + "loss": 0.0971, "step": 4820 }, { - "epoch": 0.27, - "learning_rate": 1.4656005330668e-05, - "loss": 0.1392, + "epoch": 0.54, + "learning_rate": 1.4655708573967128e-05, + "loss": 0.0931, "step": 4825 }, { - "epoch": 0.27, - "learning_rate": 1.4650452551502027e-05, - "loss": 0.0488, + "epoch": 0.54, + "learning_rate": 1.4650155486450469e-05, + "loss": 0.0386, "step": 4830 }, { - "epoch": 0.27, - "learning_rate": 1.4644899772336056e-05, - "loss": 0.1133, + "epoch": 0.54, + "learning_rate": 1.464460239893381e-05, + "loss": 0.0818, "step": 4835 }, { - "epoch": 0.27, - "learning_rate": 1.4639346993170083e-05, - "loss": 0.142, + "epoch": 0.54, + "learning_rate": 1.4639049311417149e-05, + "loss": 0.122, "step": 4840 }, { - "epoch": 0.27, - "learning_rate": 1.4633794214004111e-05, - "loss": 0.1548, + "epoch": 0.54, + "learning_rate": 1.463349622390049e-05, + "loss": 0.1686, "step": 4845 }, { - "epoch": 0.27, - "learning_rate": 1.4628241434838137e-05, - "loss": 0.328, + "epoch": 0.54, + "learning_rate": 1.4627943136383831e-05, + "loss": 0.3814, "step": 4850 }, { - "epoch": 0.27, - "learning_rate": 1.4622688655672165e-05, - "loss": 0.1151, + "epoch": 0.54, + "learning_rate": 1.4622390048867172e-05, + "loss": 0.1258, "step": 4855 }, { - "epoch": 0.27, - "learning_rate": 1.4617135876506192e-05, - "loss": 0.1361, + "epoch": 0.54, + "learning_rate": 1.4616836961350512e-05, + "loss": 0.0801, "step": 4860 }, { - "epoch": 0.27, - "learning_rate": 1.4611583097340221e-05, - "loss": 0.1218, + "epoch": 0.54, + "learning_rate": 1.4611283873833853e-05, + "loss": 0.1104, "step": 4865 }, { - "epoch": 0.27, - "learning_rate": 1.4606030318174246e-05, - "loss": 0.2385, + "epoch": 0.54, + "learning_rate": 1.4605730786317194e-05, + "loss": 0.1128, "step": 4870 }, { - "epoch": 0.27, - "learning_rate": 1.4600477539008275e-05, - "loss": 0.1267, + "epoch": 0.54, + "learning_rate": 1.4600177698800535e-05, + "loss": 0.0878, "step": 4875 }, { - "epoch": 0.27, - "learning_rate": 1.4594924759842302e-05, - "loss": 0.0889, + "epoch": 0.54, + "learning_rate": 1.4594624611283874e-05, + "loss": 0.0973, "step": 4880 }, { - "epoch": 0.27, - "learning_rate": 1.458937198067633e-05, - "loss": 0.0989, + "epoch": 0.54, + "learning_rate": 1.4589071523767215e-05, + "loss": 0.1013, "step": 4885 }, { - "epoch": 0.27, - "learning_rate": 1.4583819201510356e-05, - "loss": 0.1353, + "epoch": 0.54, + "learning_rate": 1.4583518436250556e-05, + "loss": 0.0862, "step": 4890 }, { - "epoch": 0.27, - "learning_rate": 1.4578266422344384e-05, - "loss": 0.1495, + "epoch": 0.54, + "learning_rate": 1.4577965348733897e-05, + "loss": 0.1651, "step": 4895 }, { - "epoch": 0.27, - "learning_rate": 1.4572713643178411e-05, - "loss": 0.3871, + "epoch": 0.54, + "learning_rate": 1.4572412261217237e-05, + "loss": 0.3062, "step": 4900 }, { - "epoch": 0.27, - "learning_rate": 1.456716086401244e-05, - "loss": 0.1175, + "epoch": 0.54, + "learning_rate": 1.4566859173700578e-05, + "loss": 0.0759, "step": 4905 }, { - "epoch": 0.27, - "learning_rate": 1.4561608084846465e-05, - "loss": 0.1478, + "epoch": 0.54, + "learning_rate": 1.4561306086183919e-05, + "loss": 0.085, "step": 4910 }, { - "epoch": 0.27, - "learning_rate": 1.4556055305680494e-05, - "loss": 0.1128, + "epoch": 0.55, + "learning_rate": 1.455575299866726e-05, + "loss": 0.0755, "step": 4915 }, { - "epoch": 0.27, - "learning_rate": 1.4550502526514522e-05, - "loss": 0.1267, + "epoch": 0.55, + "learning_rate": 1.45501999111506e-05, + "loss": 0.1068, "step": 4920 }, { - "epoch": 0.27, - "learning_rate": 1.454494974734855e-05, - "loss": 0.0903, + "epoch": 0.55, + "learning_rate": 1.454464682363394e-05, + "loss": 0.075, "step": 4925 }, { - "epoch": 0.27, - "learning_rate": 1.4539396968182578e-05, - "loss": 0.0873, + "epoch": 0.55, + "learning_rate": 1.4539093736117283e-05, + "loss": 0.064, "step": 4930 }, { - "epoch": 0.27, - "learning_rate": 1.4533844189016603e-05, - "loss": 0.1329, + "epoch": 0.55, + "learning_rate": 1.4533540648600624e-05, + "loss": 0.138, "step": 4935 }, { - "epoch": 0.27, - "learning_rate": 1.4528291409850632e-05, - "loss": 0.1641, + "epoch": 0.55, + "learning_rate": 1.4527987561083965e-05, + "loss": 0.0975, "step": 4940 }, { - "epoch": 0.27, - "learning_rate": 1.4522738630684659e-05, - "loss": 0.2616, + "epoch": 0.55, + "learning_rate": 1.4522434473567305e-05, + "loss": 0.1526, "step": 4945 }, { - "epoch": 0.27, - "learning_rate": 1.4517185851518688e-05, - "loss": 0.349, + "epoch": 0.55, + "learning_rate": 1.4516881386050646e-05, + "loss": 0.3765, "step": 4950 }, { - "epoch": 0.27, - "learning_rate": 1.4511633072352713e-05, - "loss": 0.1215, + "epoch": 0.55, + "learning_rate": 1.4511328298533987e-05, + "loss": 0.0871, "step": 4955 }, { - "epoch": 0.28, - "learning_rate": 1.4506080293186741e-05, - "loss": 0.1665, + "epoch": 0.55, + "learning_rate": 1.4505775211017328e-05, + "loss": 0.0829, "step": 4960 }, { - "epoch": 0.28, - "learning_rate": 1.4500527514020768e-05, - "loss": 0.1318, + "epoch": 0.55, + "learning_rate": 1.4500222123500667e-05, + "loss": 0.0882, "step": 4965 }, { - "epoch": 0.28, - "learning_rate": 1.4494974734854797e-05, - "loss": 0.1698, + "epoch": 0.55, + "learning_rate": 1.4494669035984008e-05, + "loss": 0.1186, "step": 4970 }, { - "epoch": 0.28, - "learning_rate": 1.4489421955688822e-05, - "loss": 0.1275, + "epoch": 0.55, + "learning_rate": 1.448911594846735e-05, + "loss": 0.0602, "step": 4975 }, { - "epoch": 0.28, - "learning_rate": 1.4483869176522851e-05, - "loss": 0.0977, + "epoch": 0.55, + "learning_rate": 1.448356286095069e-05, + "loss": 0.0834, "step": 4980 }, { - "epoch": 0.28, - "learning_rate": 1.4478316397356878e-05, - "loss": 0.0974, + "epoch": 0.55, + "learning_rate": 1.447800977343403e-05, + "loss": 0.0827, "step": 4985 }, { - "epoch": 0.28, - "learning_rate": 1.4472763618190907e-05, - "loss": 0.1265, + "epoch": 0.55, + "learning_rate": 1.447245668591737e-05, + "loss": 0.0868, "step": 4990 }, { - "epoch": 0.28, - "learning_rate": 1.4467210839024932e-05, - "loss": 0.1774, + "epoch": 0.55, + "learning_rate": 1.4466903598400712e-05, + "loss": 0.205, "step": 4995 }, { - "epoch": 0.28, - "learning_rate": 1.446165805985896e-05, - "loss": 0.4038, + "epoch": 0.55, + "learning_rate": 1.4461350510884053e-05, + "loss": 0.1816, "step": 5000 }, { - "epoch": 0.28, - "learning_rate": 1.4456105280692989e-05, - "loss": 0.1508, + "epoch": 0.56, + "learning_rate": 1.4455797423367392e-05, + "loss": 0.091, "step": 5005 }, { - "epoch": 0.28, - "learning_rate": 1.4450552501527016e-05, - "loss": 0.1197, + "epoch": 0.56, + "learning_rate": 1.4450244335850733e-05, + "loss": 0.1106, "step": 5010 }, { - "epoch": 0.28, - "learning_rate": 1.4444999722361045e-05, - "loss": 0.1332, + "epoch": 0.56, + "learning_rate": 1.4444691248334074e-05, + "loss": 0.0741, "step": 5015 }, { - "epoch": 0.28, - "learning_rate": 1.443944694319507e-05, - "loss": 0.1237, + "epoch": 0.56, + "learning_rate": 1.4439138160817417e-05, + "loss": 0.0894, "step": 5020 }, { - "epoch": 0.28, - "learning_rate": 1.4433894164029099e-05, - "loss": 0.0744, + "epoch": 0.56, + "learning_rate": 1.4433585073300755e-05, + "loss": 0.078, "step": 5025 }, { - "epoch": 0.28, - "learning_rate": 1.4428341384863125e-05, - "loss": 0.0844, + "epoch": 0.56, + "learning_rate": 1.4428031985784096e-05, + "loss": 0.0711, "step": 5030 }, { - "epoch": 0.28, - "learning_rate": 1.4422788605697154e-05, - "loss": 0.1149, + "epoch": 0.56, + "learning_rate": 1.4422478898267439e-05, + "loss": 0.0569, "step": 5035 }, { - "epoch": 0.28, - "learning_rate": 1.441723582653118e-05, - "loss": 0.1413, + "epoch": 0.56, + "learning_rate": 1.441692581075078e-05, + "loss": 0.132, "step": 5040 }, { - "epoch": 0.28, - "learning_rate": 1.4411683047365208e-05, - "loss": 0.2095, + "epoch": 0.56, + "learning_rate": 1.4411372723234119e-05, + "loss": 0.2066, "step": 5045 }, { - "epoch": 0.28, - "learning_rate": 1.4406130268199235e-05, - "loss": 0.4478, + "epoch": 0.56, + "learning_rate": 1.440581963571746e-05, + "loss": 0.3894, "step": 5050 }, { - "epoch": 0.28, - "learning_rate": 1.4400577489033264e-05, - "loss": 0.0956, + "epoch": 0.56, + "learning_rate": 1.4400266548200801e-05, + "loss": 0.074, "step": 5055 }, { - "epoch": 0.28, - "learning_rate": 1.4395024709867289e-05, - "loss": 0.0906, + "epoch": 0.56, + "learning_rate": 1.4394713460684142e-05, + "loss": 0.0821, "step": 5060 }, { - "epoch": 0.28, - "learning_rate": 1.4389471930701318e-05, - "loss": 0.1486, + "epoch": 0.56, + "learning_rate": 1.4389160373167483e-05, + "loss": 0.0926, "step": 5065 }, { - "epoch": 0.28, - "learning_rate": 1.4383919151535344e-05, - "loss": 0.1463, + "epoch": 0.56, + "learning_rate": 1.4383607285650823e-05, + "loss": 0.1133, "step": 5070 }, { - "epoch": 0.28, - "learning_rate": 1.4378366372369373e-05, - "loss": 0.097, + "epoch": 0.56, + "learning_rate": 1.4378054198134164e-05, + "loss": 0.0966, "step": 5075 }, { - "epoch": 0.28, - "learning_rate": 1.4372813593203398e-05, - "loss": 0.1473, + "epoch": 0.56, + "learning_rate": 1.4372501110617505e-05, + "loss": 0.0571, "step": 5080 }, { - "epoch": 0.28, - "learning_rate": 1.4367260814037427e-05, - "loss": 0.0715, + "epoch": 0.56, + "learning_rate": 1.4366948023100846e-05, + "loss": 0.0918, "step": 5085 }, { - "epoch": 0.28, - "learning_rate": 1.4361708034871454e-05, - "loss": 0.1474, + "epoch": 0.56, + "learning_rate": 1.4361394935584185e-05, + "loss": 0.1036, "step": 5090 }, { - "epoch": 0.28, - "learning_rate": 1.4356155255705481e-05, - "loss": 0.2964, + "epoch": 0.57, + "learning_rate": 1.4355841848067526e-05, + "loss": 0.1529, "step": 5095 }, { - "epoch": 0.28, - "learning_rate": 1.435060247653951e-05, - "loss": 0.413, + "epoch": 0.57, + "learning_rate": 1.4350288760550867e-05, + "loss": 0.3466, "step": 5100 }, { - "epoch": 0.28, - "learning_rate": 1.4345049697373536e-05, - "loss": 0.1322, + "epoch": 0.57, + "learning_rate": 1.4344735673034208e-05, + "loss": 0.0792, "step": 5105 }, { - "epoch": 0.28, - "learning_rate": 1.4339496918207565e-05, - "loss": 0.1631, + "epoch": 0.57, + "learning_rate": 1.4339182585517548e-05, + "loss": 0.0851, "step": 5110 }, { - "epoch": 0.28, - "learning_rate": 1.433394413904159e-05, - "loss": 0.1553, + "epoch": 0.57, + "learning_rate": 1.4333629498000889e-05, + "loss": 0.0791, "step": 5115 }, { - "epoch": 0.28, - "learning_rate": 1.4328391359875619e-05, - "loss": 0.1648, + "epoch": 0.57, + "learning_rate": 1.432807641048423e-05, + "loss": 0.1158, "step": 5120 }, { - "epoch": 0.28, - "learning_rate": 1.4322838580709646e-05, - "loss": 0.0924, + "epoch": 0.57, + "learning_rate": 1.4322523322967573e-05, + "loss": 0.0901, "step": 5125 }, { - "epoch": 0.28, - "learning_rate": 1.4317285801543675e-05, - "loss": 0.0861, + "epoch": 0.57, + "learning_rate": 1.431697023545091e-05, + "loss": 0.0801, "step": 5130 }, { - "epoch": 0.28, - "learning_rate": 1.43117330223777e-05, - "loss": 0.1427, + "epoch": 0.57, + "learning_rate": 1.4311417147934253e-05, + "loss": 0.1065, "step": 5135 }, { - "epoch": 0.29, - "learning_rate": 1.4306180243211729e-05, - "loss": 0.1756, + "epoch": 0.57, + "learning_rate": 1.4305864060417594e-05, + "loss": 0.0967, "step": 5140 }, { - "epoch": 0.29, - "learning_rate": 1.4300627464045755e-05, - "loss": 0.1543, + "epoch": 0.57, + "learning_rate": 1.4300310972900935e-05, + "loss": 0.1442, "step": 5145 }, { - "epoch": 0.29, - "learning_rate": 1.4295074684879784e-05, - "loss": 0.4378, + "epoch": 0.57, + "learning_rate": 1.4294757885384275e-05, + "loss": 0.3527, "step": 5150 }, { - "epoch": 0.29, - "learning_rate": 1.428952190571381e-05, - "loss": 0.1248, + "epoch": 0.57, + "learning_rate": 1.4289204797867616e-05, + "loss": 0.1082, "step": 5155 }, { - "epoch": 0.29, - "learning_rate": 1.4283969126547838e-05, - "loss": 0.2213, + "epoch": 0.57, + "learning_rate": 1.4283651710350957e-05, + "loss": 0.0742, "step": 5160 }, { - "epoch": 0.29, - "learning_rate": 1.4278416347381865e-05, - "loss": 0.1313, + "epoch": 0.57, + "learning_rate": 1.4278098622834298e-05, + "loss": 0.081, "step": 5165 }, { - "epoch": 0.29, - "learning_rate": 1.4272863568215894e-05, - "loss": 0.1982, + "epoch": 0.57, + "learning_rate": 1.4272545535317637e-05, + "loss": 0.1064, "step": 5170 }, { - "epoch": 0.29, - "learning_rate": 1.4267310789049919e-05, - "loss": 0.0641, + "epoch": 0.57, + "learning_rate": 1.4266992447800978e-05, + "loss": 0.1017, "step": 5175 }, { - "epoch": 0.29, - "learning_rate": 1.4261758009883948e-05, - "loss": 0.1154, + "epoch": 0.57, + "learning_rate": 1.426143936028432e-05, + "loss": 0.0786, "step": 5180 }, { - "epoch": 0.29, - "learning_rate": 1.4256205230717976e-05, - "loss": 0.1065, + "epoch": 0.58, + "learning_rate": 1.425588627276766e-05, + "loss": 0.0571, "step": 5185 }, { - "epoch": 0.29, - "learning_rate": 1.4250652451552003e-05, - "loss": 0.159, + "epoch": 0.58, + "learning_rate": 1.4250333185251e-05, + "loss": 0.0997, "step": 5190 }, { - "epoch": 0.29, - "learning_rate": 1.4245099672386032e-05, - "loss": 0.2399, + "epoch": 0.58, + "learning_rate": 1.424478009773434e-05, + "loss": 0.1396, "step": 5195 }, { - "epoch": 0.29, - "learning_rate": 1.4239546893220057e-05, - "loss": 0.3279, + "epoch": 0.58, + "learning_rate": 1.4239227010217682e-05, + "loss": 0.343, "step": 5200 }, { - "epoch": 0.29, - "learning_rate": 1.4233994114054086e-05, - "loss": 0.1197, + "epoch": 0.58, + "learning_rate": 1.4233673922701023e-05, + "loss": 0.0666, "step": 5205 }, { - "epoch": 0.29, - "learning_rate": 1.4228441334888113e-05, - "loss": 0.1623, + "epoch": 0.58, + "learning_rate": 1.4228120835184364e-05, + "loss": 0.0859, "step": 5210 }, { - "epoch": 0.29, - "learning_rate": 1.4222888555722141e-05, - "loss": 0.1429, + "epoch": 0.58, + "learning_rate": 1.4222567747667703e-05, + "loss": 0.0949, "step": 5215 }, { - "epoch": 0.29, - "learning_rate": 1.4217335776556166e-05, - "loss": 0.1794, + "epoch": 0.58, + "learning_rate": 1.4217014660151044e-05, + "loss": 0.1246, "step": 5220 }, { - "epoch": 0.29, - "learning_rate": 1.4211782997390195e-05, - "loss": 0.122, + "epoch": 0.58, + "learning_rate": 1.4211461572634385e-05, + "loss": 0.062, "step": 5225 }, { - "epoch": 0.29, - "learning_rate": 1.4206230218224222e-05, - "loss": 0.0675, + "epoch": 0.58, + "learning_rate": 1.4205908485117728e-05, + "loss": 0.0696, "step": 5230 }, { - "epoch": 0.29, - "learning_rate": 1.420067743905825e-05, - "loss": 0.0877, + "epoch": 0.58, + "learning_rate": 1.4200355397601066e-05, + "loss": 0.1217, "step": 5235 }, { - "epoch": 0.29, - "learning_rate": 1.4195124659892276e-05, - "loss": 0.1254, + "epoch": 0.58, + "learning_rate": 1.4194802310084409e-05, + "loss": 0.1105, "step": 5240 }, { - "epoch": 0.29, - "learning_rate": 1.4189571880726305e-05, - "loss": 0.262, + "epoch": 0.58, + "learning_rate": 1.418924922256775e-05, + "loss": 0.1159, "step": 5245 }, { - "epoch": 0.29, - "learning_rate": 1.4184019101560332e-05, - "loss": 0.3971, + "epoch": 0.58, + "learning_rate": 1.418369613505109e-05, + "loss": 0.2965, "step": 5250 }, { - "epoch": 0.29, - "learning_rate": 1.417846632239436e-05, - "loss": 0.1035, + "epoch": 0.58, + "learning_rate": 1.417814304753443e-05, + "loss": 0.0827, "step": 5255 }, { - "epoch": 0.29, - "learning_rate": 1.4172913543228385e-05, - "loss": 0.161, + "epoch": 0.58, + "learning_rate": 1.4172589960017771e-05, + "loss": 0.0646, "step": 5260 }, { - "epoch": 0.29, - "learning_rate": 1.4167360764062414e-05, - "loss": 0.1258, + "epoch": 0.58, + "learning_rate": 1.4167036872501112e-05, + "loss": 0.0942, "step": 5265 }, { - "epoch": 0.29, - "learning_rate": 1.4161807984896443e-05, - "loss": 0.1289, + "epoch": 0.58, + "learning_rate": 1.4161483784984453e-05, + "loss": 0.0923, "step": 5270 }, { - "epoch": 0.29, - "learning_rate": 1.415625520573047e-05, - "loss": 0.1793, + "epoch": 0.59, + "learning_rate": 1.4155930697467793e-05, + "loss": 0.0579, "step": 5275 }, { - "epoch": 0.29, - "learning_rate": 1.4150702426564498e-05, - "loss": 0.0464, + "epoch": 0.59, + "learning_rate": 1.4150377609951134e-05, + "loss": 0.0786, "step": 5280 }, { - "epoch": 0.29, - "learning_rate": 1.4145149647398524e-05, - "loss": 0.06, + "epoch": 0.59, + "learning_rate": 1.4144824522434475e-05, + "loss": 0.086, "step": 5285 }, { - "epoch": 0.29, - "learning_rate": 1.4139596868232552e-05, - "loss": 0.0949, + "epoch": 0.59, + "learning_rate": 1.4139271434917816e-05, + "loss": 0.1091, "step": 5290 }, { - "epoch": 0.29, - "learning_rate": 1.413404408906658e-05, - "loss": 0.3333, + "epoch": 0.59, + "learning_rate": 1.4133718347401155e-05, + "loss": 0.2407, "step": 5295 }, { - "epoch": 0.29, - "learning_rate": 1.4128491309900608e-05, - "loss": 0.3766, + "epoch": 0.59, + "learning_rate": 1.4128165259884496e-05, + "loss": 0.3401, "step": 5300 }, { - "epoch": 0.29, - "learning_rate": 1.4122938530734633e-05, - "loss": 0.1129, + "epoch": 0.59, + "learning_rate": 1.4122612172367837e-05, + "loss": 0.1018, "step": 5305 }, { - "epoch": 0.29, - "learning_rate": 1.4117385751568662e-05, - "loss": 0.2192, + "epoch": 0.59, + "learning_rate": 1.4117059084851178e-05, + "loss": 0.0991, "step": 5310 }, { - "epoch": 0.29, - "learning_rate": 1.4111832972402689e-05, - "loss": 0.1022, + "epoch": 0.59, + "learning_rate": 1.4111505997334518e-05, + "loss": 0.1278, "step": 5315 }, { - "epoch": 0.3, - "learning_rate": 1.4106280193236717e-05, - "loss": 0.1586, + "epoch": 0.59, + "learning_rate": 1.4105952909817859e-05, + "loss": 0.1081, "step": 5320 }, { - "epoch": 0.3, - "learning_rate": 1.4100727414070743e-05, - "loss": 0.1845, + "epoch": 0.59, + "learning_rate": 1.41003998223012e-05, + "loss": 0.1059, "step": 5325 }, { - "epoch": 0.3, - "learning_rate": 1.4095174634904771e-05, - "loss": 0.0635, + "epoch": 0.59, + "learning_rate": 1.4094846734784543e-05, + "loss": 0.0501, "step": 5330 }, { - "epoch": 0.3, - "learning_rate": 1.4089621855738798e-05, - "loss": 0.1086, + "epoch": 0.59, + "learning_rate": 1.4089293647267884e-05, + "loss": 0.1164, "step": 5335 }, { - "epoch": 0.3, - "learning_rate": 1.4084069076572827e-05, - "loss": 0.1765, + "epoch": 0.59, + "learning_rate": 1.4083740559751221e-05, + "loss": 0.1149, "step": 5340 }, { - "epoch": 0.3, - "learning_rate": 1.4078516297406852e-05, - "loss": 0.1716, + "epoch": 0.59, + "learning_rate": 1.4078187472234564e-05, + "loss": 0.1648, "step": 5345 }, { - "epoch": 0.3, - "learning_rate": 1.407296351824088e-05, - "loss": 0.421, + "epoch": 0.59, + "learning_rate": 1.4072634384717905e-05, + "loss": 0.3748, "step": 5350 }, { - "epoch": 0.3, - "learning_rate": 1.4067410739074908e-05, - "loss": 0.1385, + "epoch": 0.59, + "learning_rate": 1.4067081297201246e-05, + "loss": 0.079, "step": 5355 }, { - "epoch": 0.3, - "learning_rate": 1.4061857959908936e-05, - "loss": 0.1279, + "epoch": 0.59, + "learning_rate": 1.4061528209684586e-05, + "loss": 0.0679, "step": 5360 }, { - "epoch": 0.3, - "learning_rate": 1.4056305180742963e-05, - "loss": 0.1533, + "epoch": 0.6, + "learning_rate": 1.4055975122167927e-05, + "loss": 0.0832, "step": 5365 }, { - "epoch": 0.3, - "learning_rate": 1.405075240157699e-05, - "loss": 0.1462, + "epoch": 0.6, + "learning_rate": 1.4050422034651268e-05, + "loss": 0.114, "step": 5370 }, { - "epoch": 0.3, - "learning_rate": 1.4045199622411019e-05, - "loss": 0.0971, + "epoch": 0.6, + "learning_rate": 1.4044868947134609e-05, + "loss": 0.0807, "step": 5375 }, { - "epoch": 0.3, - "learning_rate": 1.4039646843245046e-05, - "loss": 0.042, + "epoch": 0.6, + "learning_rate": 1.4039315859617948e-05, + "loss": 0.0415, "step": 5380 }, { - "epoch": 0.3, - "learning_rate": 1.4034094064079073e-05, - "loss": 0.1182, + "epoch": 0.6, + "learning_rate": 1.403376277210129e-05, + "loss": 0.0369, "step": 5385 }, { - "epoch": 0.3, - "learning_rate": 1.40285412849131e-05, - "loss": 0.2003, + "epoch": 0.6, + "learning_rate": 1.402820968458463e-05, + "loss": 0.1212, "step": 5390 }, { - "epoch": 0.3, - "learning_rate": 1.4022988505747128e-05, - "loss": 0.2137, + "epoch": 0.6, + "learning_rate": 1.4022656597067971e-05, + "loss": 0.2648, "step": 5395 }, { - "epoch": 0.3, - "learning_rate": 1.4017435726581155e-05, - "loss": 0.2354, + "epoch": 0.6, + "learning_rate": 1.401710350955131e-05, + "loss": 0.326, "step": 5400 }, { - "epoch": 0.3, - "learning_rate": 1.4011882947415182e-05, - "loss": 0.0852, + "epoch": 0.6, + "learning_rate": 1.4011550422034652e-05, + "loss": 0.0696, "step": 5405 }, { - "epoch": 0.3, - "learning_rate": 1.400633016824921e-05, - "loss": 0.1373, + "epoch": 0.6, + "learning_rate": 1.4005997334517993e-05, + "loss": 0.091, "step": 5410 }, { - "epoch": 0.3, - "learning_rate": 1.4000777389083238e-05, - "loss": 0.1523, + "epoch": 0.6, + "learning_rate": 1.4000444247001334e-05, + "loss": 0.1021, "step": 5415 }, { - "epoch": 0.3, - "learning_rate": 1.3995224609917265e-05, - "loss": 0.1223, + "epoch": 0.6, + "learning_rate": 1.3994891159484673e-05, + "loss": 0.1057, "step": 5420 }, { - "epoch": 0.3, - "learning_rate": 1.3989671830751292e-05, - "loss": 0.1148, + "epoch": 0.6, + "learning_rate": 1.3989338071968014e-05, + "loss": 0.0742, "step": 5425 }, { - "epoch": 0.3, - "learning_rate": 1.3984119051585319e-05, - "loss": 0.0886, + "epoch": 0.6, + "learning_rate": 1.3983784984451355e-05, + "loss": 0.0442, "step": 5430 }, { - "epoch": 0.3, - "learning_rate": 1.3978566272419347e-05, - "loss": 0.1005, + "epoch": 0.6, + "learning_rate": 1.3978231896934698e-05, + "loss": 0.0694, "step": 5435 }, { - "epoch": 0.3, - "learning_rate": 1.3973013493253374e-05, - "loss": 0.1204, + "epoch": 0.6, + "learning_rate": 1.3972678809418036e-05, + "loss": 0.0808, "step": 5440 }, { - "epoch": 0.3, - "learning_rate": 1.3967460714087401e-05, - "loss": 0.2358, + "epoch": 0.6, + "learning_rate": 1.3967125721901379e-05, + "loss": 0.1707, "step": 5445 }, { - "epoch": 0.3, - "learning_rate": 1.396190793492143e-05, - "loss": 0.4988, + "epoch": 0.6, + "learning_rate": 1.396157263438472e-05, + "loss": 0.2175, "step": 5450 }, { - "epoch": 0.3, - "learning_rate": 1.3956355155755457e-05, - "loss": 0.1602, + "epoch": 0.61, + "learning_rate": 1.395601954686806e-05, + "loss": 0.0772, "step": 5455 }, { - "epoch": 0.3, - "learning_rate": 1.3950802376589485e-05, - "loss": 0.1185, + "epoch": 0.61, + "learning_rate": 1.39504664593514e-05, + "loss": 0.0992, "step": 5460 }, { - "epoch": 0.3, - "learning_rate": 1.394524959742351e-05, - "loss": 0.1606, + "epoch": 0.61, + "learning_rate": 1.3944913371834741e-05, + "loss": 0.0719, "step": 5465 }, { - "epoch": 0.3, - "learning_rate": 1.393969681825754e-05, - "loss": 0.1215, + "epoch": 0.61, + "learning_rate": 1.3939360284318082e-05, + "loss": 0.0915, "step": 5470 }, { - "epoch": 0.3, - "learning_rate": 1.3934144039091566e-05, - "loss": 0.1018, + "epoch": 0.61, + "learning_rate": 1.3933807196801423e-05, + "loss": 0.0908, "step": 5475 }, { - "epoch": 0.3, - "learning_rate": 1.3928591259925595e-05, - "loss": 0.1001, + "epoch": 0.61, + "learning_rate": 1.3928254109284764e-05, + "loss": 0.0371, "step": 5480 }, { - "epoch": 0.3, - "learning_rate": 1.392303848075962e-05, - "loss": 0.1734, + "epoch": 0.61, + "learning_rate": 1.3922701021768104e-05, + "loss": 0.0494, "step": 5485 }, { - "epoch": 0.3, - "learning_rate": 1.3917485701593649e-05, - "loss": 0.1073, + "epoch": 0.61, + "learning_rate": 1.3917147934251445e-05, + "loss": 0.1084, "step": 5490 }, { - "epoch": 0.3, - "learning_rate": 1.3911932922427676e-05, - "loss": 0.2344, + "epoch": 0.61, + "learning_rate": 1.3911594846734786e-05, + "loss": 0.14, "step": 5495 }, { - "epoch": 0.31, - "learning_rate": 1.3906380143261704e-05, - "loss": 0.4557, + "epoch": 0.61, + "learning_rate": 1.3906041759218127e-05, + "loss": 0.3693, "step": 5500 }, { - "epoch": 0.31, - "learning_rate": 1.390082736409573e-05, - "loss": 0.1253, + "epoch": 0.61, + "learning_rate": 1.3900488671701466e-05, + "loss": 0.0609, "step": 5505 }, { - "epoch": 0.31, - "learning_rate": 1.3895274584929758e-05, - "loss": 0.0939, + "epoch": 0.61, + "learning_rate": 1.3894935584184807e-05, + "loss": 0.1027, "step": 5510 }, { - "epoch": 0.31, - "learning_rate": 1.3889721805763785e-05, - "loss": 0.0968, + "epoch": 0.61, + "learning_rate": 1.3889382496668148e-05, + "loss": 0.088, "step": 5515 }, { - "epoch": 0.31, - "learning_rate": 1.3884169026597814e-05, - "loss": 0.1495, + "epoch": 0.61, + "learning_rate": 1.388382940915149e-05, + "loss": 0.1229, "step": 5520 }, { - "epoch": 0.31, - "learning_rate": 1.387861624743184e-05, - "loss": 0.1208, + "epoch": 0.61, + "learning_rate": 1.3878276321634829e-05, + "loss": 0.0449, "step": 5525 }, { - "epoch": 0.31, - "learning_rate": 1.3873063468265868e-05, - "loss": 0.0922, + "epoch": 0.61, + "learning_rate": 1.387272323411817e-05, + "loss": 0.0523, "step": 5530 }, { - "epoch": 0.31, - "learning_rate": 1.3867510689099896e-05, - "loss": 0.1613, + "epoch": 0.61, + "learning_rate": 1.3867170146601511e-05, + "loss": 0.0875, "step": 5535 }, { - "epoch": 0.31, - "learning_rate": 1.3861957909933923e-05, - "loss": 0.17, + "epoch": 0.61, + "learning_rate": 1.3861617059084854e-05, + "loss": 0.058, "step": 5540 }, { - "epoch": 0.31, - "learning_rate": 1.3856405130767952e-05, - "loss": 0.2356, + "epoch": 0.62, + "learning_rate": 1.3856063971568191e-05, + "loss": 0.1768, "step": 5545 }, { - "epoch": 0.31, - "learning_rate": 1.3850852351601977e-05, - "loss": 0.2989, + "epoch": 0.62, + "learning_rate": 1.3850510884051534e-05, + "loss": 0.3254, "step": 5550 }, { - "epoch": 0.31, - "learning_rate": 1.3845299572436006e-05, - "loss": 0.1318, + "epoch": 0.62, + "learning_rate": 1.3844957796534875e-05, + "loss": 0.0697, "step": 5555 }, { - "epoch": 0.31, - "learning_rate": 1.3839746793270033e-05, - "loss": 0.124, + "epoch": 0.62, + "learning_rate": 1.3839404709018216e-05, + "loss": 0.0821, "step": 5560 }, { - "epoch": 0.31, - "learning_rate": 1.3834194014104062e-05, - "loss": 0.0854, + "epoch": 0.62, + "learning_rate": 1.3833851621501556e-05, + "loss": 0.0736, "step": 5565 }, { - "epoch": 0.31, - "learning_rate": 1.3828641234938087e-05, - "loss": 0.1651, + "epoch": 0.62, + "learning_rate": 1.3828298533984897e-05, + "loss": 0.0821, "step": 5570 }, { - "epoch": 0.31, - "learning_rate": 1.3823088455772115e-05, - "loss": 0.1333, + "epoch": 0.62, + "learning_rate": 1.3822745446468238e-05, + "loss": 0.1, "step": 5575 }, { - "epoch": 0.31, - "learning_rate": 1.3817535676606142e-05, - "loss": 0.0837, + "epoch": 0.62, + "learning_rate": 1.3817192358951579e-05, + "loss": 0.0325, "step": 5580 }, { - "epoch": 0.31, - "learning_rate": 1.3811982897440171e-05, - "loss": 0.0767, + "epoch": 0.62, + "learning_rate": 1.3811639271434918e-05, + "loss": 0.0469, "step": 5585 }, { - "epoch": 0.31, - "learning_rate": 1.3806430118274196e-05, - "loss": 0.1533, + "epoch": 0.62, + "learning_rate": 1.380608618391826e-05, + "loss": 0.1042, "step": 5590 }, { - "epoch": 0.31, - "learning_rate": 1.3800877339108225e-05, - "loss": 0.2498, + "epoch": 0.62, + "learning_rate": 1.38005330964016e-05, + "loss": 0.1995, "step": 5595 }, { - "epoch": 0.31, - "learning_rate": 1.3795324559942252e-05, - "loss": 0.4231, + "epoch": 0.62, + "learning_rate": 1.3794980008884941e-05, + "loss": 0.1968, "step": 5600 }, { - "epoch": 0.31, - "learning_rate": 1.378977178077628e-05, - "loss": 0.1283, + "epoch": 0.62, + "learning_rate": 1.3789426921368283e-05, + "loss": 0.0811, "step": 5605 }, { - "epoch": 0.31, - "learning_rate": 1.3784219001610306e-05, - "loss": 0.1786, + "epoch": 0.62, + "learning_rate": 1.3783873833851622e-05, + "loss": 0.0966, "step": 5610 }, { - "epoch": 0.31, - "learning_rate": 1.3778666222444334e-05, - "loss": 0.1347, + "epoch": 0.62, + "learning_rate": 1.3778320746334963e-05, + "loss": 0.1066, "step": 5615 }, { - "epoch": 0.31, - "learning_rate": 1.3773113443278361e-05, - "loss": 0.1545, + "epoch": 0.62, + "learning_rate": 1.3772767658818304e-05, + "loss": 0.097, "step": 5620 }, { - "epoch": 0.31, - "learning_rate": 1.376756066411239e-05, - "loss": 0.0802, + "epoch": 0.62, + "learning_rate": 1.3767214571301645e-05, + "loss": 0.0723, "step": 5625 }, { - "epoch": 0.31, - "learning_rate": 1.3762007884946419e-05, - "loss": 0.0819, + "epoch": 0.62, + "learning_rate": 1.3761661483784984e-05, + "loss": 0.0836, "step": 5630 }, { - "epoch": 0.31, - "learning_rate": 1.3756455105780444e-05, - "loss": 0.1432, + "epoch": 0.63, + "learning_rate": 1.3756108396268325e-05, + "loss": 0.1059, "step": 5635 }, { - "epoch": 0.31, - "learning_rate": 1.3750902326614473e-05, - "loss": 0.2026, + "epoch": 0.63, + "learning_rate": 1.3750555308751668e-05, + "loss": 0.096, "step": 5640 }, { - "epoch": 0.31, - "learning_rate": 1.37453495474485e-05, - "loss": 0.1433, + "epoch": 0.63, + "learning_rate": 1.374500222123501e-05, + "loss": 0.1821, "step": 5645 }, { - "epoch": 0.31, - "learning_rate": 1.3739796768282528e-05, - "loss": 0.2551, + "epoch": 0.63, + "learning_rate": 1.3739449133718347e-05, + "loss": 0.2849, "step": 5650 }, { - "epoch": 0.31, - "learning_rate": 1.3734243989116553e-05, - "loss": 0.0918, + "epoch": 0.63, + "learning_rate": 1.373389604620169e-05, + "loss": 0.0872, "step": 5655 }, { - "epoch": 0.31, - "learning_rate": 1.3728691209950582e-05, - "loss": 0.1158, + "epoch": 0.63, + "learning_rate": 1.372834295868503e-05, + "loss": 0.0803, "step": 5660 }, { - "epoch": 0.31, - "learning_rate": 1.3723138430784609e-05, - "loss": 0.1182, + "epoch": 0.63, + "learning_rate": 1.3722789871168372e-05, + "loss": 0.088, "step": 5665 }, { - "epoch": 0.31, - "learning_rate": 1.3717585651618636e-05, - "loss": 0.1909, + "epoch": 0.63, + "learning_rate": 1.3717236783651711e-05, + "loss": 0.0868, "step": 5670 }, { - "epoch": 0.31, - "learning_rate": 1.3712032872452663e-05, - "loss": 0.0936, + "epoch": 0.63, + "learning_rate": 1.3711683696135052e-05, + "loss": 0.0899, "step": 5675 }, { - "epoch": 0.32, - "learning_rate": 1.3706480093286692e-05, - "loss": 0.0657, + "epoch": 0.63, + "learning_rate": 1.3706130608618393e-05, + "loss": 0.0617, "step": 5680 }, { - "epoch": 0.32, - "learning_rate": 1.3700927314120718e-05, - "loss": 0.1143, + "epoch": 0.63, + "learning_rate": 1.3700577521101734e-05, + "loss": 0.0903, "step": 5685 }, { - "epoch": 0.32, - "learning_rate": 1.3695374534954745e-05, - "loss": 0.1713, + "epoch": 0.63, + "learning_rate": 1.3695024433585074e-05, + "loss": 0.1025, "step": 5690 }, { - "epoch": 0.32, - "learning_rate": 1.3689821755788772e-05, - "loss": 0.1719, + "epoch": 0.63, + "learning_rate": 1.3689471346068415e-05, + "loss": 0.1936, "step": 5695 }, { - "epoch": 0.32, - "learning_rate": 1.3684268976622801e-05, - "loss": 0.3216, + "epoch": 0.63, + "learning_rate": 1.3683918258551756e-05, + "loss": 0.3353, "step": 5700 }, { - "epoch": 0.32, - "learning_rate": 1.3678716197456828e-05, - "loss": 0.1155, + "epoch": 0.63, + "learning_rate": 1.3678365171035097e-05, + "loss": 0.0707, "step": 5705 }, { - "epoch": 0.32, - "learning_rate": 1.3673163418290855e-05, - "loss": 0.137, + "epoch": 0.63, + "learning_rate": 1.3672812083518436e-05, + "loss": 0.0823, "step": 5710 }, { - "epoch": 0.32, - "learning_rate": 1.3667610639124884e-05, - "loss": 0.1279, + "epoch": 0.63, + "learning_rate": 1.3667258996001777e-05, + "loss": 0.0963, "step": 5715 }, { - "epoch": 0.32, - "learning_rate": 1.366205785995891e-05, - "loss": 0.155, + "epoch": 0.63, + "learning_rate": 1.3661705908485118e-05, + "loss": 0.1187, "step": 5720 }, { - "epoch": 0.32, - "learning_rate": 1.3656505080792939e-05, - "loss": 0.0796, + "epoch": 0.64, + "learning_rate": 1.365615282096846e-05, + "loss": 0.0632, "step": 5725 }, { - "epoch": 0.32, - "learning_rate": 1.3650952301626964e-05, - "loss": 0.1341, + "epoch": 0.64, + "learning_rate": 1.3650599733451799e-05, + "loss": 0.0524, "step": 5730 }, { - "epoch": 0.32, - "learning_rate": 1.3645399522460993e-05, - "loss": 0.1202, + "epoch": 0.64, + "learning_rate": 1.364504664593514e-05, + "loss": 0.055, "step": 5735 }, { - "epoch": 0.32, - "learning_rate": 1.363984674329502e-05, - "loss": 0.1482, + "epoch": 0.64, + "learning_rate": 1.3639493558418481e-05, + "loss": 0.149, "step": 5740 }, { - "epoch": 0.32, - "learning_rate": 1.3634293964129049e-05, - "loss": 0.1491, + "epoch": 0.64, + "learning_rate": 1.3633940470901824e-05, + "loss": 0.1794, "step": 5745 }, { - "epoch": 0.32, - "learning_rate": 1.3628741184963074e-05, - "loss": 0.4083, + "epoch": 0.64, + "learning_rate": 1.3628387383385165e-05, + "loss": 0.3633, "step": 5750 }, { - "epoch": 0.32, - "learning_rate": 1.3623188405797103e-05, - "loss": 0.1304, + "epoch": 0.64, + "learning_rate": 1.3622834295868504e-05, + "loss": 0.0882, "step": 5755 }, { - "epoch": 0.32, - "learning_rate": 1.361763562663113e-05, - "loss": 0.107, + "epoch": 0.64, + "learning_rate": 1.3617281208351845e-05, + "loss": 0.0851, "step": 5760 }, { - "epoch": 0.32, - "learning_rate": 1.3612082847465158e-05, - "loss": 0.1267, + "epoch": 0.64, + "learning_rate": 1.3611728120835186e-05, + "loss": 0.0925, "step": 5765 }, { - "epoch": 0.32, - "learning_rate": 1.3606530068299183e-05, - "loss": 0.1501, + "epoch": 0.64, + "learning_rate": 1.3606175033318527e-05, + "loss": 0.0751, "step": 5770 }, { - "epoch": 0.32, - "learning_rate": 1.3600977289133212e-05, - "loss": 0.1406, + "epoch": 0.64, + "learning_rate": 1.3600621945801867e-05, + "loss": 0.0654, "step": 5775 }, { - "epoch": 0.32, - "learning_rate": 1.3595424509967239e-05, - "loss": 0.1293, + "epoch": 0.64, + "learning_rate": 1.3595068858285208e-05, + "loss": 0.045, "step": 5780 }, { - "epoch": 0.32, - "learning_rate": 1.3589871730801268e-05, - "loss": 0.0639, + "epoch": 0.64, + "learning_rate": 1.3589515770768549e-05, + "loss": 0.0613, "step": 5785 }, { - "epoch": 0.32, - "learning_rate": 1.3584318951635293e-05, - "loss": 0.1766, + "epoch": 0.64, + "learning_rate": 1.358396268325189e-05, + "loss": 0.1491, "step": 5790 }, { - "epoch": 0.32, - "learning_rate": 1.3578766172469321e-05, - "loss": 0.2708, + "epoch": 0.64, + "learning_rate": 1.357840959573523e-05, + "loss": 0.1382, "step": 5795 }, { - "epoch": 0.32, - "learning_rate": 1.357321339330335e-05, - "loss": 0.3687, + "epoch": 0.64, + "learning_rate": 1.357285650821857e-05, + "loss": 0.2285, "step": 5800 }, { - "epoch": 0.32, - "learning_rate": 1.3567660614137377e-05, - "loss": 0.0955, + "epoch": 0.64, + "learning_rate": 1.3567303420701911e-05, + "loss": 0.1152, "step": 5805 }, { - "epoch": 0.32, - "learning_rate": 1.3562107834971406e-05, - "loss": 0.1279, + "epoch": 0.64, + "learning_rate": 1.3561750333185253e-05, + "loss": 0.0615, "step": 5810 }, { - "epoch": 0.32, - "learning_rate": 1.3556555055805431e-05, - "loss": 0.2061, + "epoch": 0.65, + "learning_rate": 1.3556197245668592e-05, + "loss": 0.0983, "step": 5815 }, { - "epoch": 0.32, - "learning_rate": 1.355100227663946e-05, - "loss": 0.1576, + "epoch": 0.65, + "learning_rate": 1.3550644158151933e-05, + "loss": 0.0896, "step": 5820 }, { - "epoch": 0.32, - "learning_rate": 1.3545449497473487e-05, - "loss": 0.1423, + "epoch": 0.65, + "learning_rate": 1.3545091070635274e-05, + "loss": 0.08, "step": 5825 }, { - "epoch": 0.32, - "learning_rate": 1.3539896718307515e-05, - "loss": 0.0753, + "epoch": 0.65, + "learning_rate": 1.3539537983118615e-05, + "loss": 0.048, "step": 5830 }, { - "epoch": 0.32, - "learning_rate": 1.353434393914154e-05, - "loss": 0.1259, + "epoch": 0.65, + "learning_rate": 1.3533984895601954e-05, + "loss": 0.0762, "step": 5835 }, { - "epoch": 0.32, - "learning_rate": 1.3528791159975569e-05, - "loss": 0.1489, + "epoch": 0.65, + "learning_rate": 1.3528431808085295e-05, + "loss": 0.0981, "step": 5840 }, { - "epoch": 0.32, - "learning_rate": 1.3523238380809596e-05, - "loss": 0.2299, + "epoch": 0.65, + "learning_rate": 1.3522878720568637e-05, + "loss": 0.1538, "step": 5845 }, { - "epoch": 0.32, - "learning_rate": 1.3517685601643625e-05, - "loss": 0.3736, + "epoch": 0.65, + "learning_rate": 1.351732563305198e-05, + "loss": 0.3178, "step": 5850 }, { - "epoch": 0.32, - "learning_rate": 1.351213282247765e-05, - "loss": 0.1262, + "epoch": 0.65, + "learning_rate": 1.3511772545535317e-05, + "loss": 0.0936, "step": 5855 }, { - "epoch": 0.33, - "learning_rate": 1.3506580043311679e-05, - "loss": 0.1038, + "epoch": 0.65, + "learning_rate": 1.350621945801866e-05, + "loss": 0.0786, "step": 5860 }, { - "epoch": 0.33, - "learning_rate": 1.3501027264145706e-05, - "loss": 0.1186, + "epoch": 0.65, + "learning_rate": 1.3500666370502e-05, + "loss": 0.0838, "step": 5865 }, { - "epoch": 0.33, - "learning_rate": 1.3495474484979734e-05, - "loss": 0.149, + "epoch": 0.65, + "learning_rate": 1.3495113282985342e-05, + "loss": 0.0975, "step": 5870 }, { - "epoch": 0.33, - "learning_rate": 1.348992170581376e-05, - "loss": 0.1261, + "epoch": 0.65, + "learning_rate": 1.3489560195468683e-05, + "loss": 0.0908, "step": 5875 }, { - "epoch": 0.33, - "learning_rate": 1.3484368926647788e-05, - "loss": 0.0345, + "epoch": 0.65, + "learning_rate": 1.3484007107952022e-05, + "loss": 0.0591, "step": 5880 }, { - "epoch": 0.33, - "learning_rate": 1.3478816147481815e-05, - "loss": 0.0758, + "epoch": 0.65, + "learning_rate": 1.3478454020435363e-05, + "loss": 0.058, "step": 5885 }, { - "epoch": 0.33, - "learning_rate": 1.3473263368315844e-05, - "loss": 0.2386, + "epoch": 0.65, + "learning_rate": 1.3472900932918704e-05, + "loss": 0.1014, "step": 5890 }, { - "epoch": 0.33, - "learning_rate": 1.3467710589149872e-05, - "loss": 0.2376, + "epoch": 0.65, + "learning_rate": 1.3467347845402045e-05, + "loss": 0.1578, "step": 5895 }, { - "epoch": 0.33, - "learning_rate": 1.3462157809983898e-05, - "loss": 0.4231, + "epoch": 0.65, + "learning_rate": 1.3461794757885385e-05, + "loss": 0.2478, "step": 5900 }, { - "epoch": 0.33, - "learning_rate": 1.3456605030817926e-05, - "loss": 0.1687, + "epoch": 0.66, + "learning_rate": 1.3456241670368726e-05, + "loss": 0.0728, "step": 5905 }, { - "epoch": 0.33, - "learning_rate": 1.3451052251651953e-05, - "loss": 0.1014, + "epoch": 0.66, + "learning_rate": 1.3450688582852067e-05, + "loss": 0.109, "step": 5910 }, { - "epoch": 0.33, - "learning_rate": 1.3445499472485982e-05, - "loss": 0.1293, + "epoch": 0.66, + "learning_rate": 1.3445135495335408e-05, + "loss": 0.091, "step": 5915 }, { - "epoch": 0.33, - "learning_rate": 1.3439946693320007e-05, - "loss": 0.132, + "epoch": 0.66, + "learning_rate": 1.3439582407818747e-05, + "loss": 0.1689, "step": 5920 }, { - "epoch": 0.33, - "learning_rate": 1.3434393914154036e-05, - "loss": 0.1038, + "epoch": 0.66, + "learning_rate": 1.3434029320302088e-05, + "loss": 0.0528, "step": 5925 }, { - "epoch": 0.33, - "learning_rate": 1.3428841134988063e-05, - "loss": 0.0456, + "epoch": 0.66, + "learning_rate": 1.342847623278543e-05, + "loss": 0.0367, "step": 5930 }, { - "epoch": 0.33, - "learning_rate": 1.3423288355822091e-05, - "loss": 0.1932, + "epoch": 0.66, + "learning_rate": 1.342292314526877e-05, + "loss": 0.1093, "step": 5935 }, { - "epoch": 0.33, - "learning_rate": 1.3417735576656117e-05, - "loss": 0.0951, + "epoch": 0.66, + "learning_rate": 1.341737005775211e-05, + "loss": 0.1172, "step": 5940 }, { - "epoch": 0.33, - "learning_rate": 1.3412182797490145e-05, - "loss": 0.1641, + "epoch": 0.66, + "learning_rate": 1.3411816970235451e-05, + "loss": 0.2165, "step": 5945 }, { - "epoch": 0.33, - "learning_rate": 1.3406630018324172e-05, - "loss": 0.3912, + "epoch": 0.66, + "learning_rate": 1.3406263882718794e-05, + "loss": 0.3271, "step": 5950 }, { - "epoch": 0.33, - "learning_rate": 1.34010772391582e-05, - "loss": 0.1788, + "epoch": 0.66, + "learning_rate": 1.3400710795202135e-05, + "loss": 0.0653, "step": 5955 }, { - "epoch": 0.33, - "learning_rate": 1.3395524459992226e-05, - "loss": 0.1008, + "epoch": 0.66, + "learning_rate": 1.3395157707685473e-05, + "loss": 0.0901, "step": 5960 }, { - "epoch": 0.33, - "learning_rate": 1.3389971680826255e-05, - "loss": 0.1296, + "epoch": 0.66, + "learning_rate": 1.3389604620168815e-05, + "loss": 0.0854, "step": 5965 }, { - "epoch": 0.33, - "learning_rate": 1.3384418901660282e-05, - "loss": 0.1362, + "epoch": 0.66, + "learning_rate": 1.3384051532652156e-05, + "loss": 0.0843, "step": 5970 }, { - "epoch": 0.33, - "learning_rate": 1.337886612249431e-05, - "loss": 0.1179, + "epoch": 0.66, + "learning_rate": 1.3378498445135497e-05, + "loss": 0.0888, "step": 5975 }, { - "epoch": 0.33, - "learning_rate": 1.3373313343328337e-05, - "loss": 0.0553, + "epoch": 0.66, + "learning_rate": 1.3372945357618837e-05, + "loss": 0.0648, "step": 5980 }, { - "epoch": 0.33, - "learning_rate": 1.3367760564162364e-05, - "loss": 0.1403, + "epoch": 0.66, + "learning_rate": 1.3367392270102178e-05, + "loss": 0.0728, "step": 5985 }, { - "epoch": 0.33, - "learning_rate": 1.3362207784996393e-05, - "loss": 0.1937, + "epoch": 0.66, + "learning_rate": 1.3361839182585519e-05, + "loss": 0.1382, "step": 5990 }, { - "epoch": 0.33, - "learning_rate": 1.335665500583042e-05, - "loss": 0.2004, + "epoch": 0.67, + "learning_rate": 1.335628609506886e-05, + "loss": 0.2024, "step": 5995 }, { - "epoch": 0.33, - "learning_rate": 1.3351102226664447e-05, - "loss": 0.5944, + "epoch": 0.67, + "learning_rate": 1.33507330075522e-05, + "loss": 0.3014, "step": 6000 }, { - "epoch": 0.33, - "eval_loss": 0.05413224548101425, - "eval_runtime": 1715.8758, - "eval_samples_per_second": 9.335, - "eval_steps_per_second": 2.334, - "eval_wer": 0.1519112874143527, + "epoch": 0.67, + "eval_loss": 0.036844413727521896, + "eval_runtime": 1820.2224, + "eval_samples_per_second": 8.8, + "eval_steps_per_second": 2.2, + "eval_wer": 0.1006130544536603, "step": 6000 }, { - "epoch": 0.33, - "learning_rate": 1.3345549447498474e-05, - "loss": 0.1627, + "epoch": 0.67, + "learning_rate": 1.334517992003554e-05, + "loss": 0.1017, "step": 6005 }, { - "epoch": 0.33, - "learning_rate": 1.3339996668332502e-05, - "loss": 0.1024, + "epoch": 0.67, + "learning_rate": 1.3339626832518881e-05, + "loss": 0.096, "step": 6010 }, { - "epoch": 0.33, - "learning_rate": 1.333444388916653e-05, - "loss": 0.1094, + "epoch": 0.67, + "learning_rate": 1.3334073745002223e-05, + "loss": 0.0874, "step": 6015 }, { - "epoch": 0.33, - "learning_rate": 1.3328891110000556e-05, - "loss": 0.137, + "epoch": 0.67, + "learning_rate": 1.3328520657485564e-05, + "loss": 0.0932, "step": 6020 }, { - "epoch": 0.33, - "learning_rate": 1.3323338330834583e-05, - "loss": 0.043, + "epoch": 0.67, + "learning_rate": 1.3322967569968903e-05, + "loss": 0.0752, "step": 6025 }, { - "epoch": 0.33, - "learning_rate": 1.3317785551668612e-05, - "loss": 0.123, + "epoch": 0.67, + "learning_rate": 1.3317414482452244e-05, + "loss": 0.0853, "step": 6030 }, { - "epoch": 0.33, - "learning_rate": 1.3312232772502637e-05, - "loss": 0.1116, + "epoch": 0.67, + "learning_rate": 1.3311861394935585e-05, + "loss": 0.0611, "step": 6035 }, { - "epoch": 0.34, - "learning_rate": 1.3306679993336666e-05, - "loss": 0.123, + "epoch": 0.67, + "learning_rate": 1.3306308307418928e-05, + "loss": 0.1506, "step": 6040 }, { - "epoch": 0.34, - "learning_rate": 1.3301127214170693e-05, - "loss": 0.2501, + "epoch": 0.67, + "learning_rate": 1.3300755219902265e-05, + "loss": 0.2175, "step": 6045 }, { - "epoch": 0.34, - "learning_rate": 1.3295574435004721e-05, - "loss": 0.2902, + "epoch": 0.67, + "learning_rate": 1.3295202132385607e-05, + "loss": 0.2922, "step": 6050 }, { - "epoch": 0.34, - "learning_rate": 1.3290021655838747e-05, - "loss": 0.0915, + "epoch": 0.67, + "learning_rate": 1.328964904486895e-05, + "loss": 0.0874, "step": 6055 }, { - "epoch": 0.34, - "learning_rate": 1.3284468876672775e-05, - "loss": 0.0852, + "epoch": 0.67, + "learning_rate": 1.328409595735229e-05, + "loss": 0.0906, "step": 6060 }, { - "epoch": 0.34, - "learning_rate": 1.3278916097506804e-05, - "loss": 0.0917, + "epoch": 0.67, + "learning_rate": 1.327854286983563e-05, + "loss": 0.0914, "step": 6065 }, { - "epoch": 0.34, - "learning_rate": 1.327336331834083e-05, - "loss": 0.1403, + "epoch": 0.67, + "learning_rate": 1.327298978231897e-05, + "loss": 0.0975, "step": 6070 }, { - "epoch": 0.34, - "learning_rate": 1.326781053917486e-05, - "loss": 0.0938, + "epoch": 0.67, + "learning_rate": 1.3267436694802312e-05, + "loss": 0.0701, "step": 6075 }, { - "epoch": 0.34, - "learning_rate": 1.3262257760008885e-05, - "loss": 0.1189, + "epoch": 0.67, + "learning_rate": 1.3261883607285653e-05, + "loss": 0.0423, "step": 6080 }, { - "epoch": 0.34, - "learning_rate": 1.3256704980842913e-05, - "loss": 0.1141, + "epoch": 0.68, + "learning_rate": 1.3256330519768992e-05, + "loss": 0.0776, "step": 6085 }, { - "epoch": 0.34, - "learning_rate": 1.325115220167694e-05, - "loss": 0.1351, + "epoch": 0.68, + "learning_rate": 1.3250777432252333e-05, + "loss": 0.1417, "step": 6090 }, { - "epoch": 0.34, - "learning_rate": 1.3245599422510969e-05, - "loss": 0.1419, + "epoch": 0.68, + "learning_rate": 1.3245224344735674e-05, + "loss": 0.1913, "step": 6095 }, { - "epoch": 0.34, - "learning_rate": 1.3240046643344994e-05, - "loss": 0.3821, + "epoch": 0.68, + "learning_rate": 1.3239671257219015e-05, + "loss": 0.3044, "step": 6100 }, { - "epoch": 0.34, - "learning_rate": 1.3234493864179023e-05, - "loss": 0.1306, + "epoch": 0.68, + "learning_rate": 1.3234118169702355e-05, + "loss": 0.0875, "step": 6105 }, { - "epoch": 0.34, - "learning_rate": 1.322894108501305e-05, - "loss": 0.1352, + "epoch": 0.68, + "learning_rate": 1.3228565082185696e-05, + "loss": 0.0929, "step": 6110 }, { - "epoch": 0.34, - "learning_rate": 1.3223388305847078e-05, - "loss": 0.1102, + "epoch": 0.68, + "learning_rate": 1.3223011994669037e-05, + "loss": 0.0918, "step": 6115 }, { - "epoch": 0.34, - "learning_rate": 1.3217835526681104e-05, - "loss": 0.1603, + "epoch": 0.68, + "learning_rate": 1.3217458907152378e-05, + "loss": 0.1143, "step": 6120 }, { - "epoch": 0.34, - "learning_rate": 1.3212282747515132e-05, - "loss": 0.1317, + "epoch": 0.68, + "learning_rate": 1.3211905819635717e-05, + "loss": 0.0624, "step": 6125 }, { - "epoch": 0.34, - "learning_rate": 1.320672996834916e-05, - "loss": 0.0713, + "epoch": 0.68, + "learning_rate": 1.3206352732119058e-05, + "loss": 0.0334, "step": 6130 }, { - "epoch": 0.34, - "learning_rate": 1.3201177189183188e-05, - "loss": 0.0511, + "epoch": 0.68, + "learning_rate": 1.32007996446024e-05, + "loss": 0.1391, "step": 6135 }, { - "epoch": 0.34, - "learning_rate": 1.3195624410017213e-05, - "loss": 0.1459, + "epoch": 0.68, + "learning_rate": 1.319524655708574e-05, + "loss": 0.1837, "step": 6140 }, { - "epoch": 0.34, - "learning_rate": 1.3190071630851242e-05, - "loss": 0.1888, + "epoch": 0.68, + "learning_rate": 1.3189693469569083e-05, + "loss": 0.1765, "step": 6145 }, { - "epoch": 0.34, - "learning_rate": 1.3184518851685269e-05, - "loss": 0.2939, + "epoch": 0.68, + "learning_rate": 1.3184140382052421e-05, + "loss": 0.3795, "step": 6150 }, { - "epoch": 0.34, - "learning_rate": 1.3178966072519297e-05, - "loss": 0.1453, + "epoch": 0.68, + "learning_rate": 1.3178587294535762e-05, + "loss": 0.0743, "step": 6155 }, { - "epoch": 0.34, - "learning_rate": 1.3173413293353326e-05, - "loss": 0.1343, + "epoch": 0.68, + "learning_rate": 1.3173034207019105e-05, + "loss": 0.0849, "step": 6160 }, { - "epoch": 0.34, - "learning_rate": 1.3167860514187351e-05, - "loss": 0.1028, + "epoch": 0.68, + "learning_rate": 1.3167481119502446e-05, + "loss": 0.0883, "step": 6165 }, { - "epoch": 0.34, - "learning_rate": 1.316230773502138e-05, - "loss": 0.1566, + "epoch": 0.68, + "learning_rate": 1.3161928031985785e-05, + "loss": 0.0848, "step": 6170 }, { - "epoch": 0.34, - "learning_rate": 1.3156754955855407e-05, - "loss": 0.1378, + "epoch": 0.69, + "learning_rate": 1.3156374944469126e-05, + "loss": 0.0605, "step": 6175 }, { - "epoch": 0.34, - "learning_rate": 1.3151202176689436e-05, - "loss": 0.1481, + "epoch": 0.69, + "learning_rate": 1.3150821856952467e-05, + "loss": 0.0624, "step": 6180 }, { - "epoch": 0.34, - "learning_rate": 1.314564939752346e-05, - "loss": 0.1071, + "epoch": 0.69, + "learning_rate": 1.3145268769435808e-05, + "loss": 0.0749, "step": 6185 }, { - "epoch": 0.34, - "learning_rate": 1.314009661835749e-05, - "loss": 0.1365, + "epoch": 0.69, + "learning_rate": 1.3139715681919148e-05, + "loss": 0.0968, "step": 6190 }, { - "epoch": 0.34, - "learning_rate": 1.3134543839191516e-05, - "loss": 0.1981, + "epoch": 0.69, + "learning_rate": 1.3134162594402489e-05, + "loss": 0.1473, "step": 6195 }, { - "epoch": 0.34, - "learning_rate": 1.3128991060025545e-05, - "loss": 0.3069, + "epoch": 0.69, + "learning_rate": 1.312860950688583e-05, + "loss": 0.3527, "step": 6200 }, { - "epoch": 0.34, - "learning_rate": 1.312343828085957e-05, - "loss": 0.1565, + "epoch": 0.69, + "learning_rate": 1.3123056419369171e-05, + "loss": 0.1292, "step": 6205 }, { - "epoch": 0.34, - "learning_rate": 1.3117885501693599e-05, - "loss": 0.1108, + "epoch": 0.69, + "learning_rate": 1.311750333185251e-05, + "loss": 0.0994, "step": 6210 }, { - "epoch": 0.34, - "learning_rate": 1.3112332722527626e-05, - "loss": 0.1099, + "epoch": 0.69, + "learning_rate": 1.3111950244335851e-05, + "loss": 0.0908, "step": 6215 }, { - "epoch": 0.35, - "learning_rate": 1.3106779943361654e-05, - "loss": 0.1057, + "epoch": 0.69, + "learning_rate": 1.3106397156819193e-05, + "loss": 0.0855, "step": 6220 }, { - "epoch": 0.35, - "learning_rate": 1.310122716419568e-05, - "loss": 0.1134, + "epoch": 0.69, + "learning_rate": 1.3100844069302534e-05, + "loss": 0.064, "step": 6225 }, { - "epoch": 0.35, - "learning_rate": 1.3095674385029708e-05, - "loss": 0.1147, + "epoch": 0.69, + "learning_rate": 1.3095290981785873e-05, + "loss": 0.0715, "step": 6230 }, { - "epoch": 0.35, - "learning_rate": 1.3090121605863735e-05, - "loss": 0.0542, + "epoch": 0.69, + "learning_rate": 1.3089737894269214e-05, + "loss": 0.0828, "step": 6235 }, { - "epoch": 0.35, - "learning_rate": 1.3084568826697764e-05, - "loss": 0.1508, + "epoch": 0.69, + "learning_rate": 1.3084184806752555e-05, + "loss": 0.0903, "step": 6240 }, { - "epoch": 0.35, - "learning_rate": 1.3079016047531791e-05, - "loss": 0.2983, + "epoch": 0.69, + "learning_rate": 1.3078631719235896e-05, + "loss": 0.1712, "step": 6245 }, { - "epoch": 0.35, - "learning_rate": 1.3073463268365818e-05, - "loss": 0.4669, + "epoch": 0.69, + "learning_rate": 1.3073078631719235e-05, + "loss": 0.2606, "step": 6250 }, { - "epoch": 0.35, - "learning_rate": 1.3067910489199847e-05, - "loss": 0.1128, + "epoch": 0.69, + "learning_rate": 1.3067525544202577e-05, + "loss": 0.1096, "step": 6255 }, { - "epoch": 0.35, - "learning_rate": 1.3062357710033873e-05, - "loss": 0.1471, + "epoch": 0.69, + "learning_rate": 1.306197245668592e-05, + "loss": 0.0771, "step": 6260 }, { - "epoch": 0.35, - "learning_rate": 1.30568049308679e-05, - "loss": 0.1082, + "epoch": 0.7, + "learning_rate": 1.305641936916926e-05, + "loss": 0.0835, "step": 6265 }, { - "epoch": 0.35, - "learning_rate": 1.3051252151701927e-05, - "loss": 0.1171, + "epoch": 0.7, + "learning_rate": 1.3050866281652598e-05, + "loss": 0.1009, "step": 6270 }, { - "epoch": 0.35, - "learning_rate": 1.3045699372535956e-05, - "loss": 0.1319, + "epoch": 0.7, + "learning_rate": 1.304531319413594e-05, + "loss": 0.0881, "step": 6275 }, { - "epoch": 0.35, - "learning_rate": 1.3040146593369983e-05, - "loss": 0.0912, + "epoch": 0.7, + "learning_rate": 1.3039760106619282e-05, + "loss": 0.0653, "step": 6280 }, { - "epoch": 0.35, - "learning_rate": 1.303459381420401e-05, - "loss": 0.1478, + "epoch": 0.7, + "learning_rate": 1.3034207019102623e-05, + "loss": 0.0751, "step": 6285 }, { - "epoch": 0.35, - "learning_rate": 1.3029041035038037e-05, - "loss": 0.1033, + "epoch": 0.7, + "learning_rate": 1.3028653931585964e-05, + "loss": 0.1077, "step": 6290 }, { - "epoch": 0.35, - "learning_rate": 1.3023488255872065e-05, - "loss": 0.1787, + "epoch": 0.7, + "learning_rate": 1.3023100844069303e-05, + "loss": 0.1567, "step": 6295 }, { - "epoch": 0.35, - "learning_rate": 1.3017935476706092e-05, - "loss": 0.4075, + "epoch": 0.7, + "learning_rate": 1.3017547756552644e-05, + "loss": 0.1985, "step": 6300 }, { - "epoch": 0.35, - "learning_rate": 1.301238269754012e-05, - "loss": 0.1128, + "epoch": 0.7, + "learning_rate": 1.3011994669035985e-05, + "loss": 0.1284, "step": 6305 }, { - "epoch": 0.35, - "learning_rate": 1.3006829918374146e-05, - "loss": 0.1184, + "epoch": 0.7, + "learning_rate": 1.3006441581519327e-05, + "loss": 0.0916, "step": 6310 }, { - "epoch": 0.35, - "learning_rate": 1.3001277139208175e-05, - "loss": 0.1095, + "epoch": 0.7, + "learning_rate": 1.3000888494002666e-05, + "loss": 0.0974, "step": 6315 }, { - "epoch": 0.35, - "learning_rate": 1.2995724360042202e-05, - "loss": 0.146, + "epoch": 0.7, + "learning_rate": 1.2995335406486007e-05, + "loss": 0.0967, "step": 6320 }, { - "epoch": 0.35, - "learning_rate": 1.2990171580876229e-05, - "loss": 0.0468, + "epoch": 0.7, + "learning_rate": 1.2989782318969348e-05, + "loss": 0.0765, "step": 6325 }, { - "epoch": 0.35, - "learning_rate": 1.2984618801710258e-05, - "loss": 0.0847, + "epoch": 0.7, + "learning_rate": 1.2984229231452689e-05, + "loss": 0.0407, "step": 6330 }, { - "epoch": 0.35, - "learning_rate": 1.2979066022544284e-05, - "loss": 0.1345, + "epoch": 0.7, + "learning_rate": 1.2978676143936028e-05, + "loss": 0.046, "step": 6335 }, { - "epoch": 0.35, - "learning_rate": 1.2973513243378313e-05, - "loss": 0.1172, + "epoch": 0.7, + "learning_rate": 1.297312305641937e-05, + "loss": 0.0735, "step": 6340 }, { - "epoch": 0.35, - "learning_rate": 1.2967960464212338e-05, - "loss": 0.3139, + "epoch": 0.7, + "learning_rate": 1.296756996890271e-05, + "loss": 0.1319, "step": 6345 }, { - "epoch": 0.35, - "learning_rate": 1.2962407685046367e-05, - "loss": 0.305, + "epoch": 0.7, + "learning_rate": 1.2962016881386053e-05, + "loss": 0.276, "step": 6350 }, { - "epoch": 0.35, - "learning_rate": 1.2956854905880394e-05, - "loss": 0.1312, + "epoch": 0.71, + "learning_rate": 1.2956463793869391e-05, + "loss": 0.0673, "step": 6355 }, { - "epoch": 0.35, - "learning_rate": 1.2951302126714423e-05, - "loss": 0.1441, + "epoch": 0.71, + "learning_rate": 1.2950910706352732e-05, + "loss": 0.0854, "step": 6360 }, { - "epoch": 0.35, - "learning_rate": 1.2945749347548448e-05, - "loss": 0.1409, + "epoch": 0.71, + "learning_rate": 1.2945357618836075e-05, + "loss": 0.0965, "step": 6365 }, { - "epoch": 0.35, - "learning_rate": 1.2940196568382477e-05, - "loss": 0.1256, + "epoch": 0.71, + "learning_rate": 1.2939804531319416e-05, + "loss": 0.0917, "step": 6370 }, { - "epoch": 0.35, - "learning_rate": 1.2934643789216503e-05, - "loss": 0.1109, + "epoch": 0.71, + "learning_rate": 1.2934251443802755e-05, + "loss": 0.0519, "step": 6375 }, { - "epoch": 0.35, - "learning_rate": 1.2929091010050532e-05, - "loss": 0.0523, + "epoch": 0.71, + "learning_rate": 1.2928698356286096e-05, + "loss": 0.0543, "step": 6380 }, { - "epoch": 0.35, - "learning_rate": 1.2923538230884557e-05, - "loss": 0.203, + "epoch": 0.71, + "learning_rate": 1.2923145268769437e-05, + "loss": 0.1, "step": 6385 }, { - "epoch": 0.35, - "learning_rate": 1.2917985451718586e-05, - "loss": 0.0992, + "epoch": 0.71, + "learning_rate": 1.2917592181252778e-05, + "loss": 0.1241, "step": 6390 }, { - "epoch": 0.35, - "learning_rate": 1.2912432672552613e-05, - "loss": 0.1871, + "epoch": 0.71, + "learning_rate": 1.2912039093736118e-05, + "loss": 0.2167, "step": 6395 }, { - "epoch": 0.36, - "learning_rate": 1.2906879893386642e-05, - "loss": 0.3306, + "epoch": 0.71, + "learning_rate": 1.2906486006219459e-05, + "loss": 0.3093, "step": 6400 }, { - "epoch": 0.36, - "learning_rate": 1.2901327114220667e-05, - "loss": 0.11, + "epoch": 0.71, + "learning_rate": 1.29009329187028e-05, + "loss": 0.0783, "step": 6405 }, { - "epoch": 0.36, - "learning_rate": 1.2895774335054695e-05, - "loss": 0.1254, + "epoch": 0.71, + "learning_rate": 1.2895379831186141e-05, + "loss": 0.0664, "step": 6410 }, { - "epoch": 0.36, - "learning_rate": 1.2890221555888722e-05, - "loss": 0.1531, + "epoch": 0.71, + "learning_rate": 1.2889826743669482e-05, + "loss": 0.0665, "step": 6415 }, { - "epoch": 0.36, - "learning_rate": 1.2884668776722751e-05, - "loss": 0.1036, + "epoch": 0.71, + "learning_rate": 1.2884273656152821e-05, + "loss": 0.0978, "step": 6420 }, { - "epoch": 0.36, - "learning_rate": 1.287911599755678e-05, - "loss": 0.0869, + "epoch": 0.71, + "learning_rate": 1.2878720568636163e-05, + "loss": 0.0608, "step": 6425 }, { - "epoch": 0.36, - "learning_rate": 1.2873563218390805e-05, - "loss": 0.085, + "epoch": 0.71, + "learning_rate": 1.2873167481119504e-05, + "loss": 0.0311, "step": 6430 }, { - "epoch": 0.36, - "learning_rate": 1.2868010439224834e-05, - "loss": 0.1312, + "epoch": 0.71, + "learning_rate": 1.2867614393602845e-05, + "loss": 0.0914, "step": 6435 }, { - "epoch": 0.36, - "learning_rate": 1.286245766005886e-05, - "loss": 0.1203, + "epoch": 0.71, + "learning_rate": 1.2862061306086184e-05, + "loss": 0.1529, "step": 6440 }, { - "epoch": 0.36, - "learning_rate": 1.285690488089289e-05, - "loss": 0.1761, + "epoch": 0.72, + "learning_rate": 1.2856508218569525e-05, + "loss": 0.1729, "step": 6445 }, { - "epoch": 0.36, - "learning_rate": 1.2851352101726914e-05, - "loss": 0.4493, + "epoch": 0.72, + "learning_rate": 1.2850955131052866e-05, + "loss": 0.2989, "step": 6450 }, { - "epoch": 0.36, - "learning_rate": 1.2845799322560943e-05, - "loss": 0.1363, + "epoch": 0.72, + "learning_rate": 1.2845402043536209e-05, + "loss": 0.0629, "step": 6455 }, { - "epoch": 0.36, - "learning_rate": 1.284024654339497e-05, - "loss": 0.1219, + "epoch": 0.72, + "learning_rate": 1.2839848956019547e-05, + "loss": 0.0793, "step": 6460 }, { - "epoch": 0.36, - "learning_rate": 1.2834693764228999e-05, - "loss": 0.112, + "epoch": 0.72, + "learning_rate": 1.2834295868502888e-05, + "loss": 0.0734, "step": 6465 }, { - "epoch": 0.36, - "learning_rate": 1.2829140985063024e-05, - "loss": 0.1142, + "epoch": 0.72, + "learning_rate": 1.282874278098623e-05, + "loss": 0.118, "step": 6470 }, { - "epoch": 0.36, - "learning_rate": 1.2823588205897053e-05, - "loss": 0.1676, + "epoch": 0.72, + "learning_rate": 1.2823189693469571e-05, + "loss": 0.0555, "step": 6475 }, { - "epoch": 0.36, - "learning_rate": 1.281803542673108e-05, - "loss": 0.0652, + "epoch": 0.72, + "learning_rate": 1.281763660595291e-05, + "loss": 0.0655, "step": 6480 }, { - "epoch": 0.36, - "learning_rate": 1.2812482647565108e-05, - "loss": 0.1131, + "epoch": 0.72, + "learning_rate": 1.2812083518436252e-05, + "loss": 0.094, "step": 6485 }, { - "epoch": 0.36, - "learning_rate": 1.2806929868399133e-05, - "loss": 0.2902, + "epoch": 0.72, + "learning_rate": 1.2806530430919593e-05, + "loss": 0.0723, "step": 6490 }, { - "epoch": 0.36, - "learning_rate": 1.2801377089233162e-05, - "loss": 0.1608, + "epoch": 0.72, + "learning_rate": 1.2800977343402934e-05, + "loss": 0.1638, "step": 6495 }, { - "epoch": 0.36, - "learning_rate": 1.2795824310067189e-05, - "loss": 0.2575, + "epoch": 0.72, + "learning_rate": 1.2795424255886273e-05, + "loss": 0.3106, "step": 6500 }, { - "epoch": 0.36, - "learning_rate": 1.2790271530901218e-05, - "loss": 0.1043, + "epoch": 0.72, + "learning_rate": 1.2789871168369614e-05, + "loss": 0.0656, "step": 6505 }, { - "epoch": 0.36, - "learning_rate": 1.2784718751735246e-05, - "loss": 0.0773, + "epoch": 0.72, + "learning_rate": 1.2784318080852955e-05, + "loss": 0.0695, "step": 6510 }, { - "epoch": 0.36, - "learning_rate": 1.2779165972569272e-05, - "loss": 0.1457, + "epoch": 0.72, + "learning_rate": 1.2778764993336297e-05, + "loss": 0.0917, "step": 6515 }, { - "epoch": 0.36, - "learning_rate": 1.27736131934033e-05, - "loss": 0.1751, + "epoch": 0.72, + "learning_rate": 1.2773211905819636e-05, + "loss": 0.1247, "step": 6520 }, { - "epoch": 0.36, - "learning_rate": 1.2768060414237327e-05, - "loss": 0.1018, + "epoch": 0.72, + "learning_rate": 1.2767658818302977e-05, + "loss": 0.0545, "step": 6525 }, { - "epoch": 0.36, - "learning_rate": 1.2762507635071356e-05, - "loss": 0.1008, + "epoch": 0.72, + "learning_rate": 1.2762105730786318e-05, + "loss": 0.0326, "step": 6530 }, { - "epoch": 0.36, - "learning_rate": 1.2756954855905381e-05, - "loss": 0.1255, + "epoch": 0.73, + "learning_rate": 1.2756552643269659e-05, + "loss": 0.0536, "step": 6535 }, { - "epoch": 0.36, - "learning_rate": 1.275140207673941e-05, - "loss": 0.1447, + "epoch": 0.73, + "learning_rate": 1.2750999555752998e-05, + "loss": 0.0613, "step": 6540 }, { - "epoch": 0.36, - "learning_rate": 1.2745849297573437e-05, - "loss": 0.1912, + "epoch": 0.73, + "learning_rate": 1.274544646823634e-05, + "loss": 0.1258, "step": 6545 }, { - "epoch": 0.36, - "learning_rate": 1.2740296518407465e-05, - "loss": 0.428, + "epoch": 0.73, + "learning_rate": 1.273989338071968e-05, + "loss": 0.266, "step": 6550 }, { - "epoch": 0.36, - "learning_rate": 1.273474373924149e-05, - "loss": 0.1376, + "epoch": 0.73, + "learning_rate": 1.2734340293203022e-05, + "loss": 0.0774, "step": 6555 }, { - "epoch": 0.36, - "learning_rate": 1.272919096007552e-05, - "loss": 0.1553, + "epoch": 0.73, + "learning_rate": 1.2728787205686364e-05, + "loss": 0.0719, "step": 6560 }, { - "epoch": 0.36, - "learning_rate": 1.2723638180909546e-05, - "loss": 0.1085, + "epoch": 0.73, + "learning_rate": 1.2723234118169702e-05, + "loss": 0.0925, "step": 6565 }, { - "epoch": 0.36, - "learning_rate": 1.2718085401743575e-05, - "loss": 0.1289, + "epoch": 0.73, + "learning_rate": 1.2717681030653045e-05, + "loss": 0.1228, "step": 6570 }, { - "epoch": 0.36, - "learning_rate": 1.27125326225776e-05, - "loss": 0.0933, + "epoch": 0.73, + "learning_rate": 1.2712127943136386e-05, + "loss": 0.1007, "step": 6575 }, { - "epoch": 0.37, - "learning_rate": 1.2706979843411629e-05, - "loss": 0.0852, + "epoch": 0.73, + "learning_rate": 1.2706574855619727e-05, + "loss": 0.0623, "step": 6580 }, { - "epoch": 0.37, - "learning_rate": 1.2701427064245656e-05, - "loss": 0.1003, + "epoch": 0.73, + "learning_rate": 1.2701021768103066e-05, + "loss": 0.0679, "step": 6585 }, { - "epoch": 0.37, - "learning_rate": 1.2695874285079684e-05, - "loss": 0.0906, + "epoch": 0.73, + "learning_rate": 1.2695468680586407e-05, + "loss": 0.1125, "step": 6590 }, { - "epoch": 0.37, - "learning_rate": 1.2690321505913711e-05, - "loss": 0.158, + "epoch": 0.73, + "learning_rate": 1.2689915593069748e-05, + "loss": 0.1594, "step": 6595 }, { - "epoch": 0.37, - "learning_rate": 1.2684768726747738e-05, - "loss": 0.3115, + "epoch": 0.73, + "learning_rate": 1.268436250555309e-05, + "loss": 0.3766, "step": 6600 }, { - "epoch": 0.37, - "learning_rate": 1.2679215947581767e-05, - "loss": 0.1225, + "epoch": 0.73, + "learning_rate": 1.2678809418036429e-05, + "loss": 0.0595, "step": 6605 }, { - "epoch": 0.37, - "learning_rate": 1.2673663168415794e-05, - "loss": 0.1325, + "epoch": 0.73, + "learning_rate": 1.267325633051977e-05, + "loss": 0.085, "step": 6610 }, { - "epoch": 0.37, - "learning_rate": 1.266811038924982e-05, - "loss": 0.1097, - "step": 6615 + "epoch": 0.73, + "learning_rate": 1.2667703243003111e-05, + "loss": 0.0736, + "step": 6615 }, { - "epoch": 0.37, - "learning_rate": 1.2662557610083848e-05, - "loss": 0.1508, + "epoch": 0.73, + "learning_rate": 1.2662150155486452e-05, + "loss": 0.0977, "step": 6620 }, { - "epoch": 0.37, - "learning_rate": 1.2657004830917876e-05, - "loss": 0.0916, + "epoch": 0.74, + "learning_rate": 1.2656597067969791e-05, + "loss": 0.0707, "step": 6625 }, { - "epoch": 0.37, - "learning_rate": 1.2651452051751902e-05, - "loss": 0.059, + "epoch": 0.74, + "learning_rate": 1.2651043980453133e-05, + "loss": 0.0653, "step": 6630 }, { - "epoch": 0.37, - "learning_rate": 1.264589927258593e-05, - "loss": 0.1261, + "epoch": 0.74, + "learning_rate": 1.2645490892936474e-05, + "loss": 0.0562, "step": 6635 }, { - "epoch": 0.37, - "learning_rate": 1.2640346493419957e-05, - "loss": 0.1421, + "epoch": 0.74, + "learning_rate": 1.2639937805419815e-05, + "loss": 0.1271, "step": 6640 }, { - "epoch": 0.37, - "learning_rate": 1.2634793714253986e-05, - "loss": 0.2639, + "epoch": 0.74, + "learning_rate": 1.2634384717903154e-05, + "loss": 0.1889, "step": 6645 }, { - "epoch": 0.37, - "learning_rate": 1.2629240935088011e-05, - "loss": 0.2909, + "epoch": 0.74, + "learning_rate": 1.2628831630386495e-05, + "loss": 0.2561, "step": 6650 }, { - "epoch": 0.37, - "learning_rate": 1.262368815592204e-05, - "loss": 0.1034, + "epoch": 0.74, + "learning_rate": 1.2623278542869836e-05, + "loss": 0.092, "step": 6655 }, { - "epoch": 0.37, - "learning_rate": 1.2618135376756067e-05, - "loss": 0.1189, + "epoch": 0.74, + "learning_rate": 1.2617725455353179e-05, + "loss": 0.0901, "step": 6660 }, { - "epoch": 0.37, - "learning_rate": 1.2612582597590095e-05, - "loss": 0.1325, + "epoch": 0.74, + "learning_rate": 1.2612172367836517e-05, + "loss": 0.0993, "step": 6665 }, { - "epoch": 0.37, - "learning_rate": 1.260702981842412e-05, - "loss": 0.1264, + "epoch": 0.74, + "learning_rate": 1.2606619280319858e-05, + "loss": 0.0945, "step": 6670 }, { - "epoch": 0.37, - "learning_rate": 1.260147703925815e-05, - "loss": 0.1649, + "epoch": 0.74, + "learning_rate": 1.26010661928032e-05, + "loss": 0.0793, "step": 6675 }, { - "epoch": 0.37, - "learning_rate": 1.2595924260092176e-05, - "loss": 0.1021, + "epoch": 0.74, + "learning_rate": 1.2595513105286541e-05, + "loss": 0.0381, "step": 6680 }, { - "epoch": 0.37, - "learning_rate": 1.2590371480926205e-05, - "loss": 0.0867, + "epoch": 0.74, + "learning_rate": 1.2589960017769883e-05, + "loss": 0.1322, "step": 6685 }, { - "epoch": 0.37, - "learning_rate": 1.2584818701760233e-05, - "loss": 0.1648, + "epoch": 0.74, + "learning_rate": 1.2584406930253222e-05, + "loss": 0.1044, "step": 6690 }, { - "epoch": 0.37, - "learning_rate": 1.2579265922594259e-05, - "loss": 0.2091, + "epoch": 0.74, + "learning_rate": 1.2578853842736563e-05, + "loss": 0.1807, "step": 6695 }, { - "epoch": 0.37, - "learning_rate": 1.2573713143428287e-05, - "loss": 0.4282, + "epoch": 0.74, + "learning_rate": 1.2573300755219904e-05, + "loss": 0.2764, "step": 6700 }, { - "epoch": 0.37, - "learning_rate": 1.2568160364262314e-05, - "loss": 0.124, + "epoch": 0.74, + "learning_rate": 1.2567747667703245e-05, + "loss": 0.0687, "step": 6705 }, { - "epoch": 0.37, - "learning_rate": 1.2562607585096343e-05, - "loss": 0.1253, + "epoch": 0.74, + "learning_rate": 1.2562194580186584e-05, + "loss": 0.0821, "step": 6710 }, { - "epoch": 0.37, - "learning_rate": 1.2557054805930368e-05, - "loss": 0.1189, + "epoch": 0.75, + "learning_rate": 1.2556641492669925e-05, + "loss": 0.0999, "step": 6715 }, { - "epoch": 0.37, - "learning_rate": 1.2551502026764397e-05, - "loss": 0.1209, + "epoch": 0.75, + "learning_rate": 1.2551088405153267e-05, + "loss": 0.0868, "step": 6720 }, { - "epoch": 0.37, - "learning_rate": 1.2545949247598424e-05, - "loss": 0.1358, + "epoch": 0.75, + "learning_rate": 1.2545535317636608e-05, + "loss": 0.0776, "step": 6725 }, { - "epoch": 0.37, - "learning_rate": 1.2540396468432452e-05, - "loss": 0.0567, + "epoch": 0.75, + "learning_rate": 1.2539982230119947e-05, + "loss": 0.0365, "step": 6730 }, { - "epoch": 0.37, - "learning_rate": 1.2534843689266478e-05, - "loss": 0.0612, + "epoch": 0.75, + "learning_rate": 1.2534429142603288e-05, + "loss": 0.0798, "step": 6735 }, { - "epoch": 0.37, - "learning_rate": 1.2529290910100506e-05, - "loss": 0.1598, + "epoch": 0.75, + "learning_rate": 1.2528876055086629e-05, + "loss": 0.0935, "step": 6740 }, { - "epoch": 0.37, - "learning_rate": 1.2523738130934533e-05, - "loss": 0.1763, + "epoch": 0.75, + "learning_rate": 1.252332296756997e-05, + "loss": 0.1345, "step": 6745 }, { - "epoch": 0.37, - "learning_rate": 1.2518185351768562e-05, - "loss": 0.3908, + "epoch": 0.75, + "learning_rate": 1.251776988005331e-05, + "loss": 0.3742, "step": 6750 }, { - "epoch": 0.37, - "learning_rate": 1.2512632572602587e-05, - "loss": 0.1011, + "epoch": 0.75, + "learning_rate": 1.251221679253665e-05, + "loss": 0.0949, "step": 6755 }, { - "epoch": 0.38, - "learning_rate": 1.2507079793436616e-05, - "loss": 0.106, + "epoch": 0.75, + "learning_rate": 1.2506663705019992e-05, + "loss": 0.0604, "step": 6760 }, { - "epoch": 0.38, - "learning_rate": 1.2501527014270643e-05, - "loss": 0.1454, + "epoch": 0.75, + "learning_rate": 1.2501110617503334e-05, + "loss": 0.0953, "step": 6765 }, { - "epoch": 0.38, - "learning_rate": 1.2495974235104671e-05, - "loss": 0.1495, + "epoch": 0.75, + "learning_rate": 1.2495557529986672e-05, + "loss": 0.1049, "step": 6770 }, { - "epoch": 0.38, - "learning_rate": 1.24904214559387e-05, - "loss": 0.1065, + "epoch": 0.75, + "learning_rate": 1.2490004442470013e-05, + "loss": 0.0469, "step": 6775 }, { - "epoch": 0.38, - "learning_rate": 1.2484868676772725e-05, - "loss": 0.065, + "epoch": 0.75, + "learning_rate": 1.2484451354953356e-05, + "loss": 0.058, "step": 6780 }, { - "epoch": 0.38, - "learning_rate": 1.2479315897606754e-05, - "loss": 0.1007, + "epoch": 0.75, + "learning_rate": 1.2478898267436697e-05, + "loss": 0.0696, "step": 6785 }, { - "epoch": 0.38, - "learning_rate": 1.2473763118440781e-05, - "loss": 0.1708, + "epoch": 0.75, + "learning_rate": 1.2473345179920036e-05, + "loss": 0.1189, "step": 6790 }, { - "epoch": 0.38, - "learning_rate": 1.246821033927481e-05, - "loss": 0.1717, + "epoch": 0.75, + "learning_rate": 1.2467792092403377e-05, + "loss": 0.2272, "step": 6795 }, { - "epoch": 0.38, - "learning_rate": 1.2462657560108835e-05, - "loss": 0.4411, + "epoch": 0.75, + "learning_rate": 1.2462239004886718e-05, + "loss": 0.3246, "step": 6800 }, { - "epoch": 0.38, - "learning_rate": 1.2457104780942863e-05, - "loss": 0.1096, + "epoch": 0.76, + "learning_rate": 1.245668591737006e-05, + "loss": 0.0933, "step": 6805 }, { - "epoch": 0.38, - "learning_rate": 1.245155200177689e-05, - "loss": 0.1293, + "epoch": 0.76, + "learning_rate": 1.2451132829853399e-05, + "loss": 0.0743, "step": 6810 }, { - "epoch": 0.38, - "learning_rate": 1.2445999222610919e-05, - "loss": 0.1362, + "epoch": 0.76, + "learning_rate": 1.244557974233674e-05, + "loss": 0.0858, "step": 6815 }, { - "epoch": 0.38, - "learning_rate": 1.2440446443444944e-05, - "loss": 0.1538, + "epoch": 0.76, + "learning_rate": 1.2440026654820081e-05, + "loss": 0.1215, "step": 6820 }, { - "epoch": 0.38, - "learning_rate": 1.2434893664278973e-05, - "loss": 0.0679, + "epoch": 0.76, + "learning_rate": 1.2434473567303422e-05, + "loss": 0.0811, "step": 6825 }, { - "epoch": 0.38, - "learning_rate": 1.2429340885113e-05, - "loss": 0.0919, + "epoch": 0.76, + "learning_rate": 1.2428920479786763e-05, + "loss": 0.0488, "step": 6830 }, { - "epoch": 0.38, - "learning_rate": 1.2423788105947028e-05, - "loss": 0.0901, + "epoch": 0.76, + "learning_rate": 1.2423367392270103e-05, + "loss": 0.0944, "step": 6835 }, { - "epoch": 0.38, - "learning_rate": 1.2418235326781054e-05, - "loss": 0.1257, + "epoch": 0.76, + "learning_rate": 1.2417814304753444e-05, + "loss": 0.0781, "step": 6840 }, { - "epoch": 0.38, - "learning_rate": 1.2412682547615082e-05, - "loss": 0.2276, + "epoch": 0.76, + "learning_rate": 1.2412261217236785e-05, + "loss": 0.207, "step": 6845 }, { - "epoch": 0.38, - "learning_rate": 1.240712976844911e-05, - "loss": 0.4954, + "epoch": 0.76, + "learning_rate": 1.2406708129720126e-05, + "loss": 0.2736, "step": 6850 }, { - "epoch": 0.38, - "learning_rate": 1.2401576989283138e-05, - "loss": 0.1201, + "epoch": 0.76, + "learning_rate": 1.2401155042203465e-05, + "loss": 0.0902, "step": 6855 }, { - "epoch": 0.38, - "learning_rate": 1.2396024210117165e-05, - "loss": 0.1194, + "epoch": 0.76, + "learning_rate": 1.2395601954686806e-05, + "loss": 0.0724, "step": 6860 }, { - "epoch": 0.38, - "learning_rate": 1.2390471430951192e-05, - "loss": 0.1221, + "epoch": 0.76, + "learning_rate": 1.2390048867170147e-05, + "loss": 0.0936, "step": 6865 }, { - "epoch": 0.38, - "learning_rate": 1.238491865178522e-05, - "loss": 0.0934, + "epoch": 0.76, + "learning_rate": 1.238449577965349e-05, + "loss": 0.1051, "step": 6870 }, { - "epoch": 0.38, - "learning_rate": 1.2379365872619247e-05, - "loss": 0.1027, + "epoch": 0.76, + "learning_rate": 1.2378942692136828e-05, + "loss": 0.0878, "step": 6875 }, { - "epoch": 0.38, - "learning_rate": 1.2373813093453274e-05, - "loss": 0.0561, + "epoch": 0.76, + "learning_rate": 1.237338960462017e-05, + "loss": 0.0677, "step": 6880 }, { - "epoch": 0.38, - "learning_rate": 1.2368260314287301e-05, - "loss": 0.1917, + "epoch": 0.76, + "learning_rate": 1.2367836517103511e-05, + "loss": 0.0564, "step": 6885 }, { - "epoch": 0.38, - "learning_rate": 1.236270753512133e-05, - "loss": 0.2151, + "epoch": 0.76, + "learning_rate": 1.2362283429586853e-05, + "loss": 0.1026, "step": 6890 }, { - "epoch": 0.38, - "learning_rate": 1.2357154755955357e-05, - "loss": 0.2768, + "epoch": 0.77, + "learning_rate": 1.2356730342070192e-05, + "loss": 0.1521, "step": 6895 }, { - "epoch": 0.38, - "learning_rate": 1.2351601976789384e-05, - "loss": 0.402, + "epoch": 0.77, + "learning_rate": 1.2351177254553533e-05, + "loss": 0.3306, "step": 6900 }, { - "epoch": 0.38, - "learning_rate": 1.234604919762341e-05, - "loss": 0.1006, + "epoch": 0.77, + "learning_rate": 1.2345624167036874e-05, + "loss": 0.0614, "step": 6905 }, { - "epoch": 0.38, - "learning_rate": 1.234049641845744e-05, - "loss": 0.1058, + "epoch": 0.77, + "learning_rate": 1.2340071079520215e-05, + "loss": 0.0764, "step": 6910 }, { - "epoch": 0.38, - "learning_rate": 1.2334943639291466e-05, - "loss": 0.0918, + "epoch": 0.77, + "learning_rate": 1.2334517992003554e-05, + "loss": 0.076, "step": 6915 }, { - "epoch": 0.38, - "learning_rate": 1.2329390860125493e-05, - "loss": 0.1459, + "epoch": 0.77, + "learning_rate": 1.2328964904486895e-05, + "loss": 0.0889, "step": 6920 }, { - "epoch": 0.38, - "learning_rate": 1.232383808095952e-05, - "loss": 0.189, + "epoch": 0.77, + "learning_rate": 1.2323411816970237e-05, + "loss": 0.0642, "step": 6925 }, { - "epoch": 0.38, - "learning_rate": 1.2318285301793549e-05, - "loss": 0.0988, + "epoch": 0.77, + "learning_rate": 1.2317858729453578e-05, + "loss": 0.0729, "step": 6930 }, { - "epoch": 0.38, - "learning_rate": 1.2312732522627576e-05, - "loss": 0.1081, + "epoch": 0.77, + "learning_rate": 1.2312305641936917e-05, + "loss": 0.0646, "step": 6935 }, { - "epoch": 0.39, - "learning_rate": 1.2307179743461603e-05, - "loss": 0.208, + "epoch": 0.77, + "learning_rate": 1.2306752554420258e-05, + "loss": 0.123, "step": 6940 }, { - "epoch": 0.39, - "learning_rate": 1.2301626964295632e-05, - "loss": 0.2445, + "epoch": 0.77, + "learning_rate": 1.2301199466903599e-05, + "loss": 0.1406, "step": 6945 }, { - "epoch": 0.39, - "learning_rate": 1.2296074185129658e-05, - "loss": 0.3171, + "epoch": 0.77, + "learning_rate": 1.229564637938694e-05, + "loss": 0.3075, "step": 6950 }, { - "epoch": 0.39, - "learning_rate": 1.2290521405963687e-05, - "loss": 0.1102, + "epoch": 0.77, + "learning_rate": 1.229009329187028e-05, + "loss": 0.0797, "step": 6955 }, { - "epoch": 0.39, - "learning_rate": 1.2284968626797712e-05, - "loss": 0.1694, + "epoch": 0.77, + "learning_rate": 1.228454020435362e-05, + "loss": 0.0726, "step": 6960 }, { - "epoch": 0.39, - "learning_rate": 1.2279415847631741e-05, - "loss": 0.1753, + "epoch": 0.77, + "learning_rate": 1.2278987116836962e-05, + "loss": 0.0699, "step": 6965 }, { - "epoch": 0.39, - "learning_rate": 1.2273863068465768e-05, - "loss": 0.1692, + "epoch": 0.77, + "learning_rate": 1.2273434029320304e-05, + "loss": 0.1097, "step": 6970 }, { - "epoch": 0.39, - "learning_rate": 1.2268310289299797e-05, - "loss": 0.1159, + "epoch": 0.77, + "learning_rate": 1.2267880941803645e-05, + "loss": 0.068, "step": 6975 }, { - "epoch": 0.39, - "learning_rate": 1.2262757510133822e-05, - "loss": 0.0865, + "epoch": 0.77, + "learning_rate": 1.2262327854286983e-05, + "loss": 0.0332, "step": 6980 }, { - "epoch": 0.39, - "learning_rate": 1.225720473096785e-05, - "loss": 0.1188, + "epoch": 0.78, + "learning_rate": 1.2256774766770326e-05, + "loss": 0.0682, "step": 6985 }, { - "epoch": 0.39, - "learning_rate": 1.2251651951801877e-05, - "loss": 0.1991, + "epoch": 0.78, + "learning_rate": 1.2251221679253667e-05, + "loss": 0.1092, "step": 6990 }, { - "epoch": 0.39, - "learning_rate": 1.2246099172635906e-05, - "loss": 0.2511, + "epoch": 0.78, + "learning_rate": 1.2245668591737008e-05, + "loss": 0.1286, "step": 6995 }, { - "epoch": 0.39, - "learning_rate": 1.2240546393469931e-05, - "loss": 0.3078, + "epoch": 0.78, + "learning_rate": 1.2240115504220347e-05, + "loss": 0.2398, "step": 7000 }, { - "epoch": 0.39, - "learning_rate": 1.223499361430396e-05, - "loss": 0.1026, + "epoch": 0.78, + "learning_rate": 1.2234562416703688e-05, + "loss": 0.1131, "step": 7005 }, { - "epoch": 0.39, - "learning_rate": 1.2229440835137987e-05, - "loss": 0.1164, + "epoch": 0.78, + "learning_rate": 1.222900932918703e-05, + "loss": 0.0692, "step": 7010 }, { - "epoch": 0.39, - "learning_rate": 1.2223888055972016e-05, - "loss": 0.1987, + "epoch": 0.78, + "learning_rate": 1.222345624167037e-05, + "loss": 0.0974, "step": 7015 }, { - "epoch": 0.39, - "learning_rate": 1.221833527680604e-05, - "loss": 0.1136, + "epoch": 0.78, + "learning_rate": 1.221790315415371e-05, + "loss": 0.0824, "step": 7020 }, { - "epoch": 0.39, - "learning_rate": 1.221278249764007e-05, - "loss": 0.0911, + "epoch": 0.78, + "learning_rate": 1.2212350066637051e-05, + "loss": 0.0727, "step": 7025 }, { - "epoch": 0.39, - "learning_rate": 1.2207229718474096e-05, - "loss": 0.0607, + "epoch": 0.78, + "learning_rate": 1.2206796979120392e-05, + "loss": 0.0549, "step": 7030 }, { - "epoch": 0.39, - "learning_rate": 1.2201676939308125e-05, - "loss": 0.1019, + "epoch": 0.78, + "learning_rate": 1.2201243891603733e-05, + "loss": 0.0778, "step": 7035 }, { - "epoch": 0.39, - "learning_rate": 1.2196124160142154e-05, - "loss": 0.1676, + "epoch": 0.78, + "learning_rate": 1.2195690804087073e-05, + "loss": 0.1629, "step": 7040 }, { - "epoch": 0.39, - "learning_rate": 1.2190571380976179e-05, - "loss": 0.2099, + "epoch": 0.78, + "learning_rate": 1.2190137716570414e-05, + "loss": 0.1378, "step": 7045 }, { - "epoch": 0.39, - "learning_rate": 1.2185018601810208e-05, - "loss": 0.3481, + "epoch": 0.78, + "learning_rate": 1.2184584629053755e-05, + "loss": 0.2265, "step": 7050 }, { - "epoch": 0.39, - "learning_rate": 1.2179465822644235e-05, - "loss": 0.1154, + "epoch": 0.78, + "learning_rate": 1.2179031541537096e-05, + "loss": 0.0768, "step": 7055 }, { - "epoch": 0.39, - "learning_rate": 1.2173913043478263e-05, - "loss": 0.1542, + "epoch": 0.78, + "learning_rate": 1.2173478454020435e-05, + "loss": 0.0784, "step": 7060 }, { - "epoch": 0.39, - "learning_rate": 1.2168360264312288e-05, - "loss": 0.1156, + "epoch": 0.78, + "learning_rate": 1.2167925366503776e-05, + "loss": 0.0943, "step": 7065 }, { - "epoch": 0.39, - "learning_rate": 1.2162807485146317e-05, - "loss": 0.1194, + "epoch": 0.78, + "learning_rate": 1.2162372278987117e-05, + "loss": 0.1106, "step": 7070 }, { - "epoch": 0.39, - "learning_rate": 1.2157254705980344e-05, - "loss": 0.1312, + "epoch": 0.79, + "learning_rate": 1.215681919147046e-05, + "loss": 0.0887, "step": 7075 }, { - "epoch": 0.39, - "learning_rate": 1.2151701926814373e-05, - "loss": 0.052, + "epoch": 0.79, + "learning_rate": 1.2151266103953798e-05, + "loss": 0.0371, "step": 7080 }, { - "epoch": 0.39, - "learning_rate": 1.2146149147648398e-05, - "loss": 0.1328, + "epoch": 0.79, + "learning_rate": 1.2145713016437139e-05, + "loss": 0.0688, "step": 7085 }, { - "epoch": 0.39, - "learning_rate": 1.2140596368482427e-05, - "loss": 0.1615, + "epoch": 0.79, + "learning_rate": 1.2140159928920481e-05, + "loss": 0.1178, "step": 7090 }, { - "epoch": 0.39, - "learning_rate": 1.2135043589316454e-05, - "loss": 0.1641, + "epoch": 0.79, + "learning_rate": 1.2134606841403823e-05, + "loss": 0.1882, "step": 7095 }, { - "epoch": 0.39, - "learning_rate": 1.2129490810150482e-05, - "loss": 0.3701, + "epoch": 0.79, + "learning_rate": 1.2129053753887164e-05, + "loss": 0.2476, "step": 7100 }, { - "epoch": 0.39, - "learning_rate": 1.2123938030984507e-05, - "loss": 0.1555, + "epoch": 0.79, + "learning_rate": 1.2123500666370503e-05, + "loss": 0.0826, "step": 7105 }, { - "epoch": 0.39, - "learning_rate": 1.2118385251818536e-05, - "loss": 0.1442, + "epoch": 0.79, + "learning_rate": 1.2117947578853844e-05, + "loss": 0.0784, "step": 7110 }, { - "epoch": 0.39, - "learning_rate": 1.2112832472652563e-05, - "loss": 0.1208, + "epoch": 0.79, + "learning_rate": 1.2112394491337185e-05, + "loss": 0.0658, "step": 7115 }, { - "epoch": 0.4, - "learning_rate": 1.2107279693486592e-05, - "loss": 0.1304, + "epoch": 0.79, + "learning_rate": 1.2106841403820526e-05, + "loss": 0.0916, "step": 7120 }, { - "epoch": 0.4, - "learning_rate": 1.210172691432062e-05, - "loss": 0.1216, + "epoch": 0.79, + "learning_rate": 1.2101288316303865e-05, + "loss": 0.076, "step": 7125 }, { - "epoch": 0.4, - "learning_rate": 1.2096174135154646e-05, - "loss": 0.0566, + "epoch": 0.79, + "learning_rate": 1.2095735228787207e-05, + "loss": 0.0296, "step": 7130 }, { - "epoch": 0.4, - "learning_rate": 1.2090621355988674e-05, - "loss": 0.0918, + "epoch": 0.79, + "learning_rate": 1.2090182141270548e-05, + "loss": 0.0875, "step": 7135 }, { - "epoch": 0.4, - "learning_rate": 1.2085068576822701e-05, - "loss": 0.1618, + "epoch": 0.79, + "learning_rate": 1.2084629053753889e-05, + "loss": 0.1144, "step": 7140 }, { - "epoch": 0.4, - "learning_rate": 1.207951579765673e-05, - "loss": 0.2134, + "epoch": 0.79, + "learning_rate": 1.2079075966237228e-05, + "loss": 0.1444, "step": 7145 }, { - "epoch": 0.4, - "learning_rate": 1.2073963018490755e-05, - "loss": 0.4406, + "epoch": 0.79, + "learning_rate": 1.2073522878720569e-05, + "loss": 0.3095, "step": 7150 }, { - "epoch": 0.4, - "learning_rate": 1.2068410239324784e-05, - "loss": 0.0804, + "epoch": 0.79, + "learning_rate": 1.206796979120391e-05, + "loss": 0.0957, "step": 7155 }, { - "epoch": 0.4, - "learning_rate": 1.206285746015881e-05, - "loss": 0.1026, + "epoch": 0.79, + "learning_rate": 1.2062416703687251e-05, + "loss": 0.0835, "step": 7160 }, { - "epoch": 0.4, - "learning_rate": 1.205730468099284e-05, - "loss": 0.1137, + "epoch": 0.8, + "learning_rate": 1.205686361617059e-05, + "loss": 0.0842, "step": 7165 }, { - "epoch": 0.4, - "learning_rate": 1.2051751901826865e-05, - "loss": 0.1262, + "epoch": 0.8, + "learning_rate": 1.2051310528653932e-05, + "loss": 0.1096, "step": 7170 }, { - "epoch": 0.4, - "learning_rate": 1.2046199122660893e-05, - "loss": 0.0953, + "epoch": 0.8, + "learning_rate": 1.2045757441137273e-05, + "loss": 0.0814, "step": 7175 }, { - "epoch": 0.4, - "learning_rate": 1.204064634349492e-05, - "loss": 0.0841, + "epoch": 0.8, + "learning_rate": 1.2040204353620615e-05, + "loss": 0.0317, "step": 7180 }, { - "epoch": 0.4, - "learning_rate": 1.2035093564328949e-05, - "loss": 0.0567, + "epoch": 0.8, + "learning_rate": 1.2034651266103953e-05, + "loss": 0.0798, "step": 7185 }, { - "epoch": 0.4, - "learning_rate": 1.2029540785162974e-05, - "loss": 0.1002, + "epoch": 0.8, + "learning_rate": 1.2029098178587296e-05, + "loss": 0.095, "step": 7190 }, { - "epoch": 0.4, - "learning_rate": 1.2023988005997003e-05, - "loss": 0.1571, + "epoch": 0.8, + "learning_rate": 1.2023545091070637e-05, + "loss": 0.103, "step": 7195 }, { - "epoch": 0.4, - "learning_rate": 1.201843522683103e-05, - "loss": 0.3455, + "epoch": 0.8, + "learning_rate": 1.2017992003553978e-05, + "loss": 0.352, "step": 7200 }, { - "epoch": 0.4, - "learning_rate": 1.2012882447665057e-05, - "loss": 0.0937, + "epoch": 0.8, + "learning_rate": 1.2012438916037317e-05, + "loss": 0.0673, "step": 7205 }, { - "epoch": 0.4, - "learning_rate": 1.2007329668499085e-05, - "loss": 0.0923, + "epoch": 0.8, + "learning_rate": 1.2006885828520658e-05, + "loss": 0.0726, "step": 7210 }, { - "epoch": 0.4, - "learning_rate": 1.2001776889333112e-05, - "loss": 0.1555, + "epoch": 0.8, + "learning_rate": 1.2001332741004e-05, + "loss": 0.0937, "step": 7215 }, { - "epoch": 0.4, - "learning_rate": 1.199622411016714e-05, - "loss": 0.1571, + "epoch": 0.8, + "learning_rate": 1.199577965348734e-05, + "loss": 0.0993, "step": 7220 }, { - "epoch": 0.4, - "learning_rate": 1.1990671331001166e-05, - "loss": 0.0866, + "epoch": 0.8, + "learning_rate": 1.199022656597068e-05, + "loss": 0.1026, "step": 7225 }, { - "epoch": 0.4, - "learning_rate": 1.1985118551835195e-05, - "loss": 0.1091, + "epoch": 0.8, + "learning_rate": 1.1984673478454021e-05, + "loss": 0.048, "step": 7230 }, { - "epoch": 0.4, - "learning_rate": 1.1979565772669222e-05, - "loss": 0.1723, + "epoch": 0.8, + "learning_rate": 1.1979120390937362e-05, + "loss": 0.0614, "step": 7235 }, { - "epoch": 0.4, - "learning_rate": 1.197401299350325e-05, - "loss": 0.0858, + "epoch": 0.8, + "learning_rate": 1.1973567303420703e-05, + "loss": 0.0894, "step": 7240 }, { - "epoch": 0.4, - "learning_rate": 1.1968460214337276e-05, - "loss": 0.1316, + "epoch": 0.8, + "learning_rate": 1.1968014215904044e-05, + "loss": 0.1637, "step": 7245 }, { - "epoch": 0.4, - "learning_rate": 1.1962907435171304e-05, - "loss": 0.5438, + "epoch": 0.8, + "learning_rate": 1.1962461128387384e-05, + "loss": 0.2663, "step": 7250 }, { - "epoch": 0.4, - "learning_rate": 1.1957354656005331e-05, - "loss": 0.1033, + "epoch": 0.81, + "learning_rate": 1.1956908040870725e-05, + "loss": 0.0688, "step": 7255 }, { - "epoch": 0.4, - "learning_rate": 1.195180187683936e-05, - "loss": 0.1173, + "epoch": 0.81, + "learning_rate": 1.1951354953354066e-05, + "loss": 0.0666, "step": 7260 }, { - "epoch": 0.4, - "learning_rate": 1.1946249097673385e-05, - "loss": 0.0717, + "epoch": 0.81, + "learning_rate": 1.1945801865837407e-05, + "loss": 0.1073, "step": 7265 }, { - "epoch": 0.4, - "learning_rate": 1.1940696318507414e-05, - "loss": 0.1282, + "epoch": 0.81, + "learning_rate": 1.1940248778320746e-05, + "loss": 0.0998, "step": 7270 }, { - "epoch": 0.4, - "learning_rate": 1.193514353934144e-05, - "loss": 0.0873, + "epoch": 0.81, + "learning_rate": 1.1934695690804087e-05, + "loss": 0.1142, "step": 7275 }, { - "epoch": 0.4, - "learning_rate": 1.192959076017547e-05, - "loss": 0.0671, + "epoch": 0.81, + "learning_rate": 1.192914260328743e-05, + "loss": 0.0357, "step": 7280 }, { - "epoch": 0.4, - "learning_rate": 1.1924037981009495e-05, - "loss": 0.1433, + "epoch": 0.81, + "learning_rate": 1.1923589515770771e-05, + "loss": 0.0909, "step": 7285 }, { - "epoch": 0.4, - "learning_rate": 1.1918485201843523e-05, - "loss": 0.1386, + "epoch": 0.81, + "learning_rate": 1.1918036428254109e-05, + "loss": 0.076, "step": 7290 }, { - "epoch": 0.4, - "learning_rate": 1.191293242267755e-05, - "loss": 0.1315, + "epoch": 0.81, + "learning_rate": 1.1912483340737451e-05, + "loss": 0.1644, "step": 7295 }, { - "epoch": 0.41, - "learning_rate": 1.1907379643511579e-05, - "loss": 0.3823, + "epoch": 0.81, + "learning_rate": 1.1906930253220793e-05, + "loss": 0.2738, "step": 7300 }, { - "epoch": 0.41, - "learning_rate": 1.1901826864345607e-05, - "loss": 0.1152, + "epoch": 0.81, + "learning_rate": 1.1901377165704134e-05, + "loss": 0.0664, "step": 7305 }, { - "epoch": 0.41, - "learning_rate": 1.1896274085179633e-05, - "loss": 0.117, + "epoch": 0.81, + "learning_rate": 1.1895824078187473e-05, + "loss": 0.0726, "step": 7310 }, { - "epoch": 0.41, - "learning_rate": 1.1890721306013661e-05, - "loss": 0.1542, + "epoch": 0.81, + "learning_rate": 1.1890270990670814e-05, + "loss": 0.1029, "step": 7315 }, { - "epoch": 0.41, - "learning_rate": 1.1885168526847688e-05, - "loss": 0.1557, + "epoch": 0.81, + "learning_rate": 1.1884717903154155e-05, + "loss": 0.0958, "step": 7320 }, { - "epoch": 0.41, - "learning_rate": 1.1879615747681717e-05, - "loss": 0.1171, + "epoch": 0.81, + "learning_rate": 1.1879164815637496e-05, + "loss": 0.0916, "step": 7325 }, { - "epoch": 0.41, - "learning_rate": 1.1874062968515742e-05, - "loss": 0.0791, + "epoch": 0.81, + "learning_rate": 1.1873611728120835e-05, + "loss": 0.038, "step": 7330 }, { - "epoch": 0.41, - "learning_rate": 1.186851018934977e-05, - "loss": 0.085, + "epoch": 0.81, + "learning_rate": 1.1868058640604177e-05, + "loss": 0.1326, "step": 7335 }, { - "epoch": 0.41, - "learning_rate": 1.1862957410183798e-05, - "loss": 0.157, + "epoch": 0.81, + "learning_rate": 1.1862505553087518e-05, + "loss": 0.1143, "step": 7340 }, { - "epoch": 0.41, - "learning_rate": 1.1857404631017826e-05, - "loss": 0.2116, + "epoch": 0.82, + "learning_rate": 1.1856952465570859e-05, + "loss": 0.1256, "step": 7345 }, { - "epoch": 0.41, - "learning_rate": 1.1851851851851852e-05, - "loss": 0.4489, + "epoch": 0.82, + "learning_rate": 1.1851399378054198e-05, + "loss": 0.2505, "step": 7350 }, { - "epoch": 0.41, - "learning_rate": 1.184629907268588e-05, - "loss": 0.1485, + "epoch": 0.82, + "learning_rate": 1.1845846290537539e-05, + "loss": 0.0835, "step": 7355 }, { - "epoch": 0.41, - "learning_rate": 1.1840746293519907e-05, - "loss": 0.097, + "epoch": 0.82, + "learning_rate": 1.184029320302088e-05, + "loss": 0.0724, "step": 7360 }, { - "epoch": 0.41, - "learning_rate": 1.1835193514353936e-05, - "loss": 0.146, + "epoch": 0.82, + "learning_rate": 1.1834740115504221e-05, + "loss": 0.0806, "step": 7365 }, { - "epoch": 0.41, - "learning_rate": 1.1829640735187961e-05, - "loss": 0.1248, + "epoch": 0.82, + "learning_rate": 1.1829187027987562e-05, + "loss": 0.103, "step": 7370 }, { - "epoch": 0.41, - "learning_rate": 1.182408795602199e-05, - "loss": 0.0745, + "epoch": 0.82, + "learning_rate": 1.1823633940470902e-05, + "loss": 0.0907, "step": 7375 }, { - "epoch": 0.41, - "learning_rate": 1.1818535176856017e-05, - "loss": 0.0444, + "epoch": 0.82, + "learning_rate": 1.1818080852954243e-05, + "loss": 0.0558, "step": 7380 }, { - "epoch": 0.41, - "learning_rate": 1.1812982397690045e-05, - "loss": 0.0466, + "epoch": 0.82, + "learning_rate": 1.1812527765437585e-05, + "loss": 0.0855, "step": 7385 }, { - "epoch": 0.41, - "learning_rate": 1.1807429618524074e-05, - "loss": 0.1801, + "epoch": 0.82, + "learning_rate": 1.1806974677920927e-05, + "loss": 0.109, "step": 7390 }, { - "epoch": 0.41, - "learning_rate": 1.18018768393581e-05, - "loss": 0.1436, + "epoch": 0.82, + "learning_rate": 1.1801421590404264e-05, + "loss": 0.1699, "step": 7395 }, { - "epoch": 0.41, - "learning_rate": 1.1796324060192128e-05, - "loss": 0.3079, + "epoch": 0.82, + "learning_rate": 1.1795868502887607e-05, + "loss": 0.1998, "step": 7400 }, { - "epoch": 0.41, - "learning_rate": 1.1790771281026155e-05, - "loss": 0.0907, + "epoch": 0.82, + "learning_rate": 1.1790315415370948e-05, + "loss": 0.0815, "step": 7405 }, { - "epoch": 0.41, - "learning_rate": 1.1785218501860183e-05, - "loss": 0.0962, + "epoch": 0.82, + "learning_rate": 1.1784762327854289e-05, + "loss": 0.0679, "step": 7410 }, { - "epoch": 0.41, - "learning_rate": 1.1779665722694209e-05, - "loss": 0.1144, + "epoch": 0.82, + "learning_rate": 1.1779209240337628e-05, + "loss": 0.0747, "step": 7415 }, { - "epoch": 0.41, - "learning_rate": 1.1774112943528237e-05, - "loss": 0.1293, + "epoch": 0.82, + "learning_rate": 1.177365615282097e-05, + "loss": 0.1136, "step": 7420 }, { - "epoch": 0.41, - "learning_rate": 1.1768560164362264e-05, - "loss": 0.0808, + "epoch": 0.82, + "learning_rate": 1.176810306530431e-05, + "loss": 0.0891, "step": 7425 }, { - "epoch": 0.41, - "learning_rate": 1.1763007385196293e-05, - "loss": 0.0684, + "epoch": 0.82, + "learning_rate": 1.1762549977787652e-05, + "loss": 0.0596, "step": 7430 }, { - "epoch": 0.41, - "learning_rate": 1.1757454606030318e-05, - "loss": 0.0765, + "epoch": 0.83, + "learning_rate": 1.1756996890270991e-05, + "loss": 0.075, "step": 7435 }, { - "epoch": 0.41, - "learning_rate": 1.1751901826864347e-05, - "loss": 0.1493, + "epoch": 0.83, + "learning_rate": 1.1751443802754332e-05, + "loss": 0.1402, "step": 7440 }, { - "epoch": 0.41, - "learning_rate": 1.1746349047698374e-05, - "loss": 0.1191, + "epoch": 0.83, + "learning_rate": 1.1745890715237673e-05, + "loss": 0.1379, "step": 7445 }, { - "epoch": 0.41, - "learning_rate": 1.1740796268532402e-05, - "loss": 0.2765, + "epoch": 0.83, + "learning_rate": 1.1740337627721014e-05, + "loss": 0.2628, "step": 7450 }, { - "epoch": 0.41, - "learning_rate": 1.1735243489366428e-05, - "loss": 0.1075, + "epoch": 0.83, + "learning_rate": 1.1734784540204354e-05, + "loss": 0.0775, "step": 7455 }, { - "epoch": 0.41, - "learning_rate": 1.1729690710200456e-05, - "loss": 0.1029, + "epoch": 0.83, + "learning_rate": 1.1729231452687695e-05, + "loss": 0.0763, "step": 7460 }, { - "epoch": 0.41, - "learning_rate": 1.1724137931034483e-05, - "loss": 0.0941, + "epoch": 0.83, + "learning_rate": 1.1723678365171036e-05, + "loss": 0.09, "step": 7465 }, { - "epoch": 0.41, - "learning_rate": 1.1718585151868512e-05, - "loss": 0.1487, + "epoch": 0.83, + "learning_rate": 1.1718125277654377e-05, + "loss": 0.1043, "step": 7470 }, { - "epoch": 0.41, - "learning_rate": 1.1713032372702539e-05, - "loss": 0.0814, + "epoch": 0.83, + "learning_rate": 1.1712572190137716e-05, + "loss": 0.0497, "step": 7475 }, { - "epoch": 0.42, - "learning_rate": 1.1707479593536566e-05, - "loss": 0.0605, + "epoch": 0.83, + "learning_rate": 1.1707019102621057e-05, + "loss": 0.0547, "step": 7480 }, { - "epoch": 0.42, - "learning_rate": 1.1701926814370594e-05, - "loss": 0.0998, + "epoch": 0.83, + "learning_rate": 1.1701466015104398e-05, + "loss": 0.0468, "step": 7485 }, { - "epoch": 0.42, - "learning_rate": 1.1696374035204621e-05, - "loss": 0.1742, + "epoch": 0.83, + "learning_rate": 1.1695912927587741e-05, + "loss": 0.0631, "step": 7490 }, { - "epoch": 0.42, - "learning_rate": 1.1690821256038648e-05, - "loss": 0.1423, + "epoch": 0.83, + "learning_rate": 1.1690359840071079e-05, + "loss": 0.1546, "step": 7495 }, { - "epoch": 0.42, - "learning_rate": 1.168637903270587e-05, - "loss": 0.4971, + "epoch": 0.83, + "learning_rate": 1.1684806752554421e-05, + "loss": 0.3041, "step": 7500 }, { - "epoch": 0.42, - "learning_rate": 1.1680826253539898e-05, - "loss": 0.0946, + "epoch": 0.83, + "learning_rate": 1.1679253665037763e-05, + "loss": 0.0867, "step": 7505 }, { - "epoch": 0.42, - "learning_rate": 1.1675273474373925e-05, - "loss": 0.1281, + "epoch": 0.83, + "learning_rate": 1.1673700577521104e-05, + "loss": 0.0911, "step": 7510 }, { - "epoch": 0.42, - "learning_rate": 1.1669720695207954e-05, - "loss": 0.1161, + "epoch": 0.83, + "learning_rate": 1.1668147490004445e-05, + "loss": 0.1018, "step": 7515 }, { - "epoch": 0.42, - "learning_rate": 1.1664167916041979e-05, - "loss": 0.1538, + "epoch": 0.83, + "learning_rate": 1.1662594402487784e-05, + "loss": 0.0885, "step": 7520 }, { - "epoch": 0.42, - "learning_rate": 1.1658615136876008e-05, - "loss": 0.1138, + "epoch": 0.84, + "learning_rate": 1.1657041314971125e-05, + "loss": 0.0657, "step": 7525 }, { - "epoch": 0.42, - "learning_rate": 1.1653062357710035e-05, - "loss": 0.08, + "epoch": 0.84, + "learning_rate": 1.1651488227454466e-05, + "loss": 0.0568, "step": 7530 }, { - "epoch": 0.42, - "learning_rate": 1.1647509578544063e-05, - "loss": 0.0917, + "epoch": 0.84, + "learning_rate": 1.1645935139937807e-05, + "loss": 0.076, "step": 7535 }, { - "epoch": 0.42, - "learning_rate": 1.1641956799378089e-05, - "loss": 0.1029, + "epoch": 0.84, + "learning_rate": 1.1640382052421147e-05, + "loss": 0.1338, "step": 7540 }, { - "epoch": 0.42, - "learning_rate": 1.1636404020212117e-05, - "loss": 0.1555, + "epoch": 0.84, + "learning_rate": 1.1634828964904488e-05, + "loss": 0.1244, "step": 7545 }, { - "epoch": 0.42, - "learning_rate": 1.1630851241046144e-05, - "loss": 0.5797, + "epoch": 0.84, + "learning_rate": 1.1629275877387829e-05, + "loss": 0.2539, "step": 7550 }, { - "epoch": 0.42, - "learning_rate": 1.1625298461880173e-05, - "loss": 0.1056, + "epoch": 0.84, + "learning_rate": 1.162372278987117e-05, + "loss": 0.0679, "step": 7555 }, { - "epoch": 0.42, - "learning_rate": 1.1619745682714198e-05, - "loss": 0.1344, + "epoch": 0.84, + "learning_rate": 1.1618169702354509e-05, + "loss": 0.0776, "step": 7560 }, { - "epoch": 0.42, - "learning_rate": 1.1614192903548227e-05, - "loss": 0.1613, + "epoch": 0.84, + "learning_rate": 1.161261661483785e-05, + "loss": 0.0873, "step": 7565 }, { - "epoch": 0.42, - "learning_rate": 1.1608640124382254e-05, - "loss": 0.1386, + "epoch": 0.84, + "learning_rate": 1.1607063527321191e-05, + "loss": 0.1074, "step": 7570 }, { - "epoch": 0.42, - "learning_rate": 1.1603087345216282e-05, - "loss": 0.0699, + "epoch": 0.84, + "learning_rate": 1.1601510439804532e-05, + "loss": 0.0508, "step": 7575 }, { - "epoch": 0.42, - "learning_rate": 1.1597534566050308e-05, - "loss": 0.0939, + "epoch": 0.84, + "learning_rate": 1.1595957352287872e-05, + "loss": 0.0702, "step": 7580 }, { - "epoch": 0.42, - "learning_rate": 1.1591981786884336e-05, - "loss": 0.1394, + "epoch": 0.84, + "learning_rate": 1.1590404264771213e-05, + "loss": 0.0716, "step": 7585 }, { - "epoch": 0.42, - "learning_rate": 1.1586429007718365e-05, - "loss": 0.1519, + "epoch": 0.84, + "learning_rate": 1.1584851177254555e-05, + "loss": 0.1019, "step": 7590 }, { - "epoch": 0.42, - "learning_rate": 1.1580876228552392e-05, - "loss": 0.2223, + "epoch": 0.84, + "learning_rate": 1.1579298089737897e-05, + "loss": 0.1635, "step": 7595 }, { - "epoch": 0.42, - "learning_rate": 1.157532344938642e-05, - "loss": 0.3872, + "epoch": 0.84, + "learning_rate": 1.1573745002221234e-05, + "loss": 0.371, "step": 7600 }, { - "epoch": 0.42, - "learning_rate": 1.1569770670220446e-05, - "loss": 0.1475, + "epoch": 0.84, + "learning_rate": 1.1568191914704577e-05, + "loss": 0.0837, "step": 7605 }, { - "epoch": 0.42, - "learning_rate": 1.1564217891054474e-05, - "loss": 0.1131, + "epoch": 0.84, + "learning_rate": 1.1562638827187918e-05, + "loss": 0.0913, "step": 7610 }, { - "epoch": 0.42, - "learning_rate": 1.1558665111888501e-05, - "loss": 0.1318, + "epoch": 0.85, + "learning_rate": 1.1557085739671259e-05, + "loss": 0.078, "step": 7615 }, { - "epoch": 0.42, - "learning_rate": 1.155311233272253e-05, - "loss": 0.0939, + "epoch": 0.85, + "learning_rate": 1.1551532652154598e-05, + "loss": 0.0948, "step": 7620 }, { - "epoch": 0.42, - "learning_rate": 1.1547559553556555e-05, - "loss": 0.1337, + "epoch": 0.85, + "learning_rate": 1.154597956463794e-05, + "loss": 0.0505, "step": 7625 }, { - "epoch": 0.42, - "learning_rate": 1.1542006774390584e-05, - "loss": 0.0527, + "epoch": 0.85, + "learning_rate": 1.154042647712128e-05, + "loss": 0.109, "step": 7630 }, { - "epoch": 0.42, - "learning_rate": 1.153645399522461e-05, - "loss": 0.1811, + "epoch": 0.85, + "learning_rate": 1.1534873389604622e-05, + "loss": 0.0851, "step": 7635 }, { - "epoch": 0.42, - "learning_rate": 1.153090121605864e-05, - "loss": 0.1872, + "epoch": 0.85, + "learning_rate": 1.1529320302087963e-05, + "loss": 0.0835, "step": 7640 }, { - "epoch": 0.42, - "learning_rate": 1.1525348436892665e-05, - "loss": 0.1479, + "epoch": 0.85, + "learning_rate": 1.1523767214571302e-05, + "loss": 0.2377, "step": 7645 }, { - "epoch": 0.42, - "learning_rate": 1.1519795657726693e-05, - "loss": 0.4209, + "epoch": 0.85, + "learning_rate": 1.1518214127054643e-05, + "loss": 0.308, "step": 7650 }, { - "epoch": 0.42, - "learning_rate": 1.151424287856072e-05, - "loss": 0.1194, + "epoch": 0.85, + "learning_rate": 1.1512661039537984e-05, + "loss": 0.0853, "step": 7655 }, { - "epoch": 0.43, - "learning_rate": 1.1508690099394749e-05, - "loss": 0.1303, + "epoch": 0.85, + "learning_rate": 1.1507107952021325e-05, + "loss": 0.0764, "step": 7660 }, { - "epoch": 0.43, - "learning_rate": 1.1503137320228774e-05, - "loss": 0.1341, + "epoch": 0.85, + "learning_rate": 1.1501554864504665e-05, + "loss": 0.1057, "step": 7665 }, { - "epoch": 0.43, - "learning_rate": 1.1497584541062803e-05, - "loss": 0.1288, + "epoch": 0.85, + "learning_rate": 1.1496001776988006e-05, + "loss": 0.0947, "step": 7670 }, { - "epoch": 0.43, - "learning_rate": 1.1492031761896831e-05, - "loss": 0.1082, + "epoch": 0.85, + "learning_rate": 1.1490448689471347e-05, + "loss": 0.0707, "step": 7675 }, { - "epoch": 0.43, - "learning_rate": 1.1486478982730858e-05, - "loss": 0.1112, + "epoch": 0.85, + "learning_rate": 1.1484895601954688e-05, + "loss": 0.0622, "step": 7680 }, { - "epoch": 0.43, - "learning_rate": 1.1480926203564885e-05, - "loss": 0.0792, + "epoch": 0.85, + "learning_rate": 1.1479342514438027e-05, + "loss": 0.0737, "step": 7685 }, { - "epoch": 0.43, - "learning_rate": 1.1475373424398912e-05, - "loss": 0.1594, + "epoch": 0.85, + "learning_rate": 1.1473789426921368e-05, + "loss": 0.1298, "step": 7690 }, { - "epoch": 0.43, - "learning_rate": 1.1469820645232941e-05, - "loss": 0.184, + "epoch": 0.85, + "learning_rate": 1.1468236339404711e-05, + "loss": 0.1445, "step": 7695 }, { - "epoch": 0.43, - "learning_rate": 1.1464267866066968e-05, - "loss": 0.2832, + "epoch": 0.85, + "learning_rate": 1.1463793869391383e-05, + "loss": 0.5195, "step": 7700 }, { - "epoch": 0.43, - "learning_rate": 1.1458715086900995e-05, - "loss": 0.132, + "epoch": 0.86, + "learning_rate": 1.1458240781874724e-05, + "loss": 0.0784, "step": 7705 }, { - "epoch": 0.43, - "learning_rate": 1.1453162307735022e-05, - "loss": 0.1174, + "epoch": 0.86, + "learning_rate": 1.1452687694358065e-05, + "loss": 0.08, "step": 7710 }, { - "epoch": 0.43, - "learning_rate": 1.144760952856905e-05, - "loss": 0.1156, + "epoch": 0.86, + "learning_rate": 1.1447134606841406e-05, + "loss": 0.112, "step": 7715 }, { - "epoch": 0.43, - "learning_rate": 1.1442056749403077e-05, - "loss": 0.1271, + "epoch": 0.86, + "learning_rate": 1.1441581519324745e-05, + "loss": 0.0875, "step": 7720 }, { - "epoch": 0.43, - "learning_rate": 1.1436503970237104e-05, - "loss": 0.0965, + "epoch": 0.86, + "learning_rate": 1.1436028431808086e-05, + "loss": 0.0986, "step": 7725 }, { - "epoch": 0.43, - "learning_rate": 1.1430951191071131e-05, - "loss": 0.0653, + "epoch": 0.86, + "learning_rate": 1.1430475344291427e-05, + "loss": 0.0193, "step": 7730 }, { - "epoch": 0.43, - "learning_rate": 1.142539841190516e-05, - "loss": 0.121, + "epoch": 0.86, + "learning_rate": 1.1424922256774768e-05, + "loss": 0.0789, "step": 7735 }, { - "epoch": 0.43, - "learning_rate": 1.1419845632739185e-05, - "loss": 0.1107, + "epoch": 0.86, + "learning_rate": 1.1419369169258108e-05, + "loss": 0.0805, "step": 7740 }, { - "epoch": 0.43, - "learning_rate": 1.1414292853573214e-05, - "loss": 0.2061, + "epoch": 0.86, + "learning_rate": 1.1413816081741449e-05, + "loss": 0.1398, "step": 7745 }, { - "epoch": 0.43, - "learning_rate": 1.140874007440724e-05, - "loss": 0.4359, + "epoch": 0.86, + "learning_rate": 1.140826299422479e-05, + "loss": 0.33, "step": 7750 }, { - "epoch": 0.43, - "learning_rate": 1.140318729524127e-05, - "loss": 0.1237, + "epoch": 0.86, + "learning_rate": 1.140270990670813e-05, + "loss": 0.0824, "step": 7755 }, { - "epoch": 0.43, - "learning_rate": 1.1397634516075298e-05, - "loss": 0.1217, + "epoch": 0.86, + "learning_rate": 1.139715681919147e-05, + "loss": 0.069, "step": 7760 }, { - "epoch": 0.43, - "learning_rate": 1.1392081736909323e-05, - "loss": 0.1232, + "epoch": 0.86, + "learning_rate": 1.1391603731674811e-05, + "loss": 0.0972, "step": 7765 }, { - "epoch": 0.43, - "learning_rate": 1.1386528957743352e-05, - "loss": 0.1375, + "epoch": 0.86, + "learning_rate": 1.1386050644158152e-05, + "loss": 0.1048, "step": 7770 }, { - "epoch": 0.43, - "learning_rate": 1.1380976178577379e-05, - "loss": 0.1595, + "epoch": 0.86, + "learning_rate": 1.1380497556641495e-05, + "loss": 0.0684, "step": 7775 }, { - "epoch": 0.43, - "learning_rate": 1.1375423399411408e-05, - "loss": 0.0878, + "epoch": 0.86, + "learning_rate": 1.1374944469124833e-05, + "loss": 0.0667, "step": 7780 }, { - "epoch": 0.43, - "learning_rate": 1.1369870620245433e-05, - "loss": 0.0618, + "epoch": 0.86, + "learning_rate": 1.1369391381608175e-05, + "loss": 0.0953, "step": 7785 }, { - "epoch": 0.43, - "learning_rate": 1.1364317841079461e-05, - "loss": 0.1696, + "epoch": 0.86, + "learning_rate": 1.1363838294091517e-05, + "loss": 0.1007, "step": 7790 }, { - "epoch": 0.43, - "learning_rate": 1.1358765061913488e-05, - "loss": 0.1687, + "epoch": 0.87, + "learning_rate": 1.1358285206574858e-05, + "loss": 0.1114, "step": 7795 }, { - "epoch": 0.43, - "learning_rate": 1.1353212282747517e-05, - "loss": 0.3339, + "epoch": 0.87, + "learning_rate": 1.1352732119058197e-05, + "loss": 0.3188, "step": 7800 }, { - "epoch": 0.43, - "learning_rate": 1.1347659503581542e-05, - "loss": 0.0886, + "epoch": 0.87, + "learning_rate": 1.1347179031541538e-05, + "loss": 0.0714, "step": 7805 }, { - "epoch": 0.43, - "learning_rate": 1.1342106724415571e-05, - "loss": 0.0751, + "epoch": 0.87, + "learning_rate": 1.1341625944024879e-05, + "loss": 0.0624, "step": 7810 }, { - "epoch": 0.43, - "learning_rate": 1.1336553945249598e-05, - "loss": 0.1337, + "epoch": 0.87, + "learning_rate": 1.133607285650822e-05, + "loss": 0.0733, "step": 7815 }, { - "epoch": 0.43, - "learning_rate": 1.1331001166083627e-05, - "loss": 0.1457, + "epoch": 0.87, + "learning_rate": 1.133051976899156e-05, + "loss": 0.1139, "step": 7820 }, { - "epoch": 0.43, - "learning_rate": 1.1325448386917652e-05, - "loss": 0.0653, + "epoch": 0.87, + "learning_rate": 1.13249666814749e-05, + "loss": 0.0523, "step": 7825 }, { - "epoch": 0.43, - "learning_rate": 1.131989560775168e-05, - "loss": 0.0806, + "epoch": 0.87, + "learning_rate": 1.1319413593958242e-05, + "loss": 0.0635, "step": 7830 }, { - "epoch": 0.43, - "learning_rate": 1.1314342828585707e-05, - "loss": 0.1549, + "epoch": 0.87, + "learning_rate": 1.1313860506441583e-05, + "loss": 0.1009, "step": 7835 }, { - "epoch": 0.44, - "learning_rate": 1.1308790049419736e-05, - "loss": 0.1529, + "epoch": 0.87, + "learning_rate": 1.1308307418924924e-05, + "loss": 0.0955, "step": 7840 }, { - "epoch": 0.44, - "learning_rate": 1.1303237270253761e-05, - "loss": 0.1298, + "epoch": 0.87, + "learning_rate": 1.1302754331408263e-05, + "loss": 0.1388, "step": 7845 }, { - "epoch": 0.44, - "learning_rate": 1.129768449108779e-05, - "loss": 0.3576, + "epoch": 0.87, + "learning_rate": 1.1297201243891604e-05, + "loss": 0.2546, "step": 7850 }, { - "epoch": 0.44, - "learning_rate": 1.1292131711921819e-05, - "loss": 0.1375, + "epoch": 0.87, + "learning_rate": 1.1291648156374945e-05, + "loss": 0.0709, "step": 7855 }, { - "epoch": 0.44, - "learning_rate": 1.1286578932755845e-05, - "loss": 0.0873, + "epoch": 0.87, + "learning_rate": 1.1286095068858286e-05, + "loss": 0.0806, "step": 7860 }, { - "epoch": 0.44, - "learning_rate": 1.1281026153589874e-05, - "loss": 0.0914, + "epoch": 0.87, + "learning_rate": 1.1280541981341626e-05, + "loss": 0.0654, "step": 7865 }, { - "epoch": 0.44, - "learning_rate": 1.12754733744239e-05, - "loss": 0.1319, + "epoch": 0.87, + "learning_rate": 1.1274988893824967e-05, + "loss": 0.0894, "step": 7870 }, { - "epoch": 0.44, - "learning_rate": 1.1269920595257928e-05, - "loss": 0.0837, + "epoch": 0.87, + "learning_rate": 1.126943580630831e-05, + "loss": 0.0977, "step": 7875 }, { - "epoch": 0.44, - "learning_rate": 1.1264367816091955e-05, - "loss": 0.0385, + "epoch": 0.87, + "learning_rate": 1.126388271879165e-05, + "loss": 0.0561, "step": 7880 }, { - "epoch": 0.44, - "learning_rate": 1.1258815036925984e-05, - "loss": 0.1033, + "epoch": 0.88, + "learning_rate": 1.1258329631274988e-05, + "loss": 0.0614, "step": 7885 }, { - "epoch": 0.44, - "learning_rate": 1.1253262257760009e-05, - "loss": 0.1939, + "epoch": 0.88, + "learning_rate": 1.1252776543758331e-05, + "loss": 0.111, "step": 7890 }, { - "epoch": 0.44, - "learning_rate": 1.1247709478594038e-05, - "loss": 0.1537, + "epoch": 0.88, + "learning_rate": 1.1247223456241672e-05, + "loss": 0.1592, "step": 7895 }, { - "epoch": 0.44, - "learning_rate": 1.1242156699428064e-05, - "loss": 0.4123, + "epoch": 0.88, + "learning_rate": 1.1241670368725013e-05, + "loss": 0.2755, "step": 7900 }, { - "epoch": 0.44, - "learning_rate": 1.1236603920262093e-05, - "loss": 0.1021, + "epoch": 0.88, + "learning_rate": 1.1236117281208353e-05, + "loss": 0.0819, "step": 7905 }, { - "epoch": 0.44, - "learning_rate": 1.1231051141096118e-05, - "loss": 0.1013, + "epoch": 0.88, + "learning_rate": 1.1230564193691694e-05, + "loss": 0.0998, "step": 7910 }, { - "epoch": 0.44, - "learning_rate": 1.1225498361930147e-05, - "loss": 0.1203, + "epoch": 0.88, + "learning_rate": 1.1225011106175035e-05, + "loss": 0.0772, "step": 7915 }, { - "epoch": 0.44, - "learning_rate": 1.1219945582764174e-05, - "loss": 0.106, + "epoch": 0.88, + "learning_rate": 1.1219458018658376e-05, + "loss": 0.0856, "step": 7920 }, { - "epoch": 0.44, - "learning_rate": 1.1214392803598203e-05, - "loss": 0.0942, + "epoch": 0.88, + "learning_rate": 1.1213904931141715e-05, + "loss": 0.0872, "step": 7925 }, { - "epoch": 0.44, - "learning_rate": 1.1208840024432228e-05, - "loss": 0.0428, + "epoch": 0.88, + "learning_rate": 1.1208351843625056e-05, + "loss": 0.038, "step": 7930 }, { - "epoch": 0.44, - "learning_rate": 1.1203287245266257e-05, - "loss": 0.0724, + "epoch": 0.88, + "learning_rate": 1.1202798756108397e-05, + "loss": 0.1082, "step": 7935 }, { - "epoch": 0.44, - "learning_rate": 1.1197734466100285e-05, - "loss": 0.1407, + "epoch": 0.88, + "learning_rate": 1.1197245668591738e-05, + "loss": 0.156, "step": 7940 }, { - "epoch": 0.44, - "learning_rate": 1.1192181686934312e-05, - "loss": 0.2703, + "epoch": 0.88, + "learning_rate": 1.1191692581075078e-05, + "loss": 0.1274, "step": 7945 }, { - "epoch": 0.44, - "learning_rate": 1.1186628907768339e-05, - "loss": 0.4318, + "epoch": 0.88, + "learning_rate": 1.1186139493558419e-05, + "loss": 0.2681, "step": 7950 }, { - "epoch": 0.44, - "learning_rate": 1.1181076128602366e-05, - "loss": 0.1065, + "epoch": 0.88, + "learning_rate": 1.118058640604176e-05, + "loss": 0.0749, "step": 7955 }, { - "epoch": 0.44, - "learning_rate": 1.1175523349436395e-05, - "loss": 0.1323, + "epoch": 0.88, + "learning_rate": 1.11750333185251e-05, + "loss": 0.0729, "step": 7960 }, { - "epoch": 0.44, - "learning_rate": 1.1169970570270422e-05, - "loss": 0.1075, + "epoch": 0.88, + "learning_rate": 1.1169480231008442e-05, + "loss": 0.0877, "step": 7965 }, { - "epoch": 0.44, - "learning_rate": 1.1164417791104449e-05, - "loss": 0.1283, + "epoch": 0.88, + "learning_rate": 1.1163927143491781e-05, + "loss": 0.0814, "step": 7970 }, { - "epoch": 0.44, - "learning_rate": 1.1158865011938475e-05, - "loss": 0.0806, + "epoch": 0.89, + "learning_rate": 1.1158374055975122e-05, + "loss": 0.0642, "step": 7975 }, { - "epoch": 0.44, - "learning_rate": 1.1153312232772504e-05, - "loss": 0.1719, + "epoch": 0.89, + "learning_rate": 1.1152820968458465e-05, + "loss": 0.0322, "step": 7980 }, { - "epoch": 0.44, - "learning_rate": 1.1147759453606531e-05, - "loss": 0.0678, + "epoch": 0.89, + "learning_rate": 1.1147267880941806e-05, + "loss": 0.0572, "step": 7985 }, { - "epoch": 0.44, - "learning_rate": 1.1142206674440558e-05, - "loss": 0.1344, + "epoch": 0.89, + "learning_rate": 1.1141714793425144e-05, + "loss": 0.104, "step": 7990 }, { - "epoch": 0.44, - "learning_rate": 1.1136653895274585e-05, - "loss": 0.2375, + "epoch": 0.89, + "learning_rate": 1.1136161705908487e-05, + "loss": 0.111, "step": 7995 }, { - "epoch": 0.44, - "learning_rate": 1.1131101116108614e-05, - "loss": 0.3759, + "epoch": 0.89, + "learning_rate": 1.1130608618391828e-05, + "loss": 0.302, "step": 8000 }, { - "epoch": 0.44, - "eval_loss": 0.04906551539897919, - "eval_runtime": 1723.037, - "eval_samples_per_second": 9.296, - "eval_steps_per_second": 2.324, - "eval_wer": 0.12998557518932563, + "epoch": 0.89, + "eval_loss": 0.03474867343902588, + "eval_runtime": 1810.8597, + "eval_samples_per_second": 8.846, + "eval_steps_per_second": 2.212, + "eval_wer": 0.08851424450054093, "step": 8000 }, { - "epoch": 0.44, - "learning_rate": 1.112554833694264e-05, - "loss": 0.0861, + "epoch": 0.89, + "learning_rate": 1.1125055530875169e-05, + "loss": 0.106, "step": 8005 }, { - "epoch": 0.44, - "learning_rate": 1.1119995557776668e-05, - "loss": 0.1191, + "epoch": 0.89, + "learning_rate": 1.1119502443358508e-05, + "loss": 0.0708, "step": 8010 }, { - "epoch": 0.44, - "learning_rate": 1.1114442778610694e-05, - "loss": 0.1263, + "epoch": 0.89, + "learning_rate": 1.1113949355841849e-05, + "loss": 0.1007, "step": 8015 }, { - "epoch": 0.45, - "learning_rate": 1.1108889999444723e-05, - "loss": 0.112, + "epoch": 0.89, + "learning_rate": 1.110839626832519e-05, + "loss": 0.0792, "step": 8020 }, { - "epoch": 0.45, - "learning_rate": 1.1103337220278752e-05, - "loss": 0.0554, + "epoch": 0.89, + "learning_rate": 1.1102843180808531e-05, + "loss": 0.1007, "step": 8025 }, { - "epoch": 0.45, - "learning_rate": 1.1097784441112777e-05, - "loss": 0.1366, + "epoch": 0.89, + "learning_rate": 1.109729009329187e-05, + "loss": 0.0455, "step": 8030 }, { - "epoch": 0.45, - "learning_rate": 1.1092231661946806e-05, - "loss": 0.1, + "epoch": 0.89, + "learning_rate": 1.1091737005775212e-05, + "loss": 0.0634, "step": 8035 }, { - "epoch": 0.45, - "learning_rate": 1.1086678882780833e-05, - "loss": 0.1335, + "epoch": 0.89, + "learning_rate": 1.1086183918258553e-05, + "loss": 0.0717, "step": 8040 }, { - "epoch": 0.45, - "learning_rate": 1.1081126103614861e-05, - "loss": 0.202, + "epoch": 0.89, + "learning_rate": 1.1080630830741894e-05, + "loss": 0.1224, "step": 8045 }, { - "epoch": 0.45, - "learning_rate": 1.1075573324448886e-05, - "loss": 0.2224, + "epoch": 0.89, + "learning_rate": 1.1075077743225233e-05, + "loss": 0.3224, "step": 8050 }, { - "epoch": 0.45, - "learning_rate": 1.1070020545282915e-05, - "loss": 0.1226, + "epoch": 0.89, + "learning_rate": 1.1069524655708574e-05, + "loss": 0.0863, "step": 8055 }, { - "epoch": 0.45, - "learning_rate": 1.1064467766116942e-05, - "loss": 0.1156, + "epoch": 0.89, + "learning_rate": 1.1063971568191915e-05, + "loss": 0.0761, "step": 8060 }, { - "epoch": 0.45, - "learning_rate": 1.105891498695097e-05, - "loss": 0.1659, + "epoch": 0.9, + "learning_rate": 1.1058418480675256e-05, + "loss": 0.1265, "step": 8065 }, { - "epoch": 0.45, - "learning_rate": 1.1053362207784996e-05, - "loss": 0.0975, + "epoch": 0.9, + "learning_rate": 1.1052865393158596e-05, + "loss": 0.1053, "step": 8070 }, { - "epoch": 0.45, - "learning_rate": 1.1047809428619025e-05, - "loss": 0.0571, + "epoch": 0.9, + "learning_rate": 1.1047312305641937e-05, + "loss": 0.0667, "step": 8075 }, { - "epoch": 0.45, - "learning_rate": 1.1042256649453052e-05, - "loss": 0.1556, + "epoch": 0.9, + "learning_rate": 1.1041759218125278e-05, + "loss": 0.0568, "step": 8080 }, { - "epoch": 0.45, - "learning_rate": 1.103670387028708e-05, - "loss": 0.1254, + "epoch": 0.9, + "learning_rate": 1.103620613060862e-05, + "loss": 0.0537, "step": 8085 }, { - "epoch": 0.45, - "learning_rate": 1.1031151091121105e-05, - "loss": 0.1396, + "epoch": 0.9, + "learning_rate": 1.1030653043091958e-05, + "loss": 0.1009, "step": 8090 }, { - "epoch": 0.45, - "learning_rate": 1.1025598311955134e-05, - "loss": 0.2073, + "epoch": 0.9, + "learning_rate": 1.1025099955575301e-05, + "loss": 0.1139, "step": 8095 }, { - "epoch": 0.45, - "learning_rate": 1.1020045532789161e-05, - "loss": 0.4153, + "epoch": 0.9, + "learning_rate": 1.1019546868058642e-05, + "loss": 0.2392, "step": 8100 }, { - "epoch": 0.45, - "learning_rate": 1.101449275362319e-05, - "loss": 0.0755, + "epoch": 0.9, + "learning_rate": 1.1013993780541983e-05, + "loss": 0.091, "step": 8105 }, { - "epoch": 0.45, - "learning_rate": 1.1008939974457215e-05, - "loss": 0.1001, + "epoch": 0.9, + "learning_rate": 1.1008440693025324e-05, + "loss": 0.0759, "step": 8110 }, { - "epoch": 0.45, - "learning_rate": 1.1003387195291244e-05, - "loss": 0.1755, - "step": 8115 + "epoch": 0.9, + "learning_rate": 1.1002887605508664e-05, + "loss": 0.0879, + "step": 8115 }, { - "epoch": 0.45, - "learning_rate": 1.0997834416125272e-05, - "loss": 0.1319, + "epoch": 0.9, + "learning_rate": 1.0997334517992005e-05, + "loss": 0.09, "step": 8120 }, { - "epoch": 0.45, - "learning_rate": 1.09922816369593e-05, - "loss": 0.1253, + "epoch": 0.9, + "learning_rate": 1.0991781430475346e-05, + "loss": 0.0958, "step": 8125 }, { - "epoch": 0.45, - "learning_rate": 1.0986728857793328e-05, - "loss": 0.1016, + "epoch": 0.9, + "learning_rate": 1.0986228342958687e-05, + "loss": 0.0568, "step": 8130 }, { - "epoch": 0.45, - "learning_rate": 1.0981176078627353e-05, - "loss": 0.0642, + "epoch": 0.9, + "learning_rate": 1.0980675255442026e-05, + "loss": 0.1016, "step": 8135 }, { - "epoch": 0.45, - "learning_rate": 1.0975623299461382e-05, - "loss": 0.1049, + "epoch": 0.9, + "learning_rate": 1.0975122167925367e-05, + "loss": 0.1496, "step": 8140 }, { - "epoch": 0.45, - "learning_rate": 1.0970070520295409e-05, - "loss": 0.1423, + "epoch": 0.9, + "learning_rate": 1.0969569080408708e-05, + "loss": 0.1523, "step": 8145 }, { - "epoch": 0.45, - "learning_rate": 1.0964517741129437e-05, - "loss": 0.3099, + "epoch": 0.9, + "learning_rate": 1.096401599289205e-05, + "loss": 0.2466, "step": 8150 }, { - "epoch": 0.45, - "learning_rate": 1.0958964961963463e-05, - "loss": 0.096, + "epoch": 0.91, + "learning_rate": 1.0958462905375389e-05, + "loss": 0.062, "step": 8155 }, { - "epoch": 0.45, - "learning_rate": 1.0953412182797491e-05, - "loss": 0.1019, + "epoch": 0.91, + "learning_rate": 1.095290981785873e-05, + "loss": 0.0759, "step": 8160 }, { - "epoch": 0.45, - "learning_rate": 1.0947859403631518e-05, - "loss": 0.125, + "epoch": 0.91, + "learning_rate": 1.094735673034207e-05, + "loss": 0.0717, "step": 8165 }, { - "epoch": 0.45, - "learning_rate": 1.0942306624465547e-05, - "loss": 0.0965, + "epoch": 0.91, + "learning_rate": 1.0941803642825412e-05, + "loss": 0.1179, "step": 8170 }, { - "epoch": 0.45, - "learning_rate": 1.0936753845299572e-05, - "loss": 0.1087, + "epoch": 0.91, + "learning_rate": 1.0936250555308751e-05, + "loss": 0.0686, "step": 8175 }, { - "epoch": 0.45, - "learning_rate": 1.09312010661336e-05, - "loss": 0.0734, + "epoch": 0.91, + "learning_rate": 1.0930697467792092e-05, + "loss": 0.031, "step": 8180 }, { - "epoch": 0.45, - "learning_rate": 1.0925648286967628e-05, - "loss": 0.1459, + "epoch": 0.91, + "learning_rate": 1.0925144380275435e-05, + "loss": 0.0653, "step": 8185 }, { - "epoch": 0.45, - "learning_rate": 1.0920095507801656e-05, - "loss": 0.1123, + "epoch": 0.91, + "learning_rate": 1.0919591292758776e-05, + "loss": 0.1195, "step": 8190 }, { - "epoch": 0.45, - "learning_rate": 1.0914542728635682e-05, - "loss": 0.1878, + "epoch": 0.91, + "learning_rate": 1.0914038205242114e-05, + "loss": 0.2072, "step": 8195 }, { - "epoch": 0.46, - "learning_rate": 1.090898994946971e-05, - "loss": 0.3884, + "epoch": 0.91, + "learning_rate": 1.0908485117725457e-05, + "loss": 0.2945, "step": 8200 }, { - "epoch": 0.46, - "learning_rate": 1.0903437170303739e-05, - "loss": 0.1347, + "epoch": 0.91, + "learning_rate": 1.0902932030208798e-05, + "loss": 0.0873, "step": 8205 }, { - "epoch": 0.46, - "learning_rate": 1.0897884391137766e-05, - "loss": 0.1257, + "epoch": 0.91, + "learning_rate": 1.0897378942692139e-05, + "loss": 0.0985, "step": 8210 }, { - "epoch": 0.46, - "learning_rate": 1.0892331611971794e-05, - "loss": 0.0914, + "epoch": 0.91, + "learning_rate": 1.0891825855175478e-05, + "loss": 0.0986, "step": 8215 }, { - "epoch": 0.46, - "learning_rate": 1.088677883280582e-05, - "loss": 0.1282, + "epoch": 0.91, + "learning_rate": 1.0886272767658819e-05, + "loss": 0.1102, "step": 8220 }, { - "epoch": 0.46, - "learning_rate": 1.0881226053639848e-05, - "loss": 0.0784, + "epoch": 0.91, + "learning_rate": 1.088071968014216e-05, + "loss": 0.075, "step": 8225 }, { - "epoch": 0.46, - "learning_rate": 1.0875673274473875e-05, - "loss": 0.0874, + "epoch": 0.91, + "learning_rate": 1.0875166592625501e-05, + "loss": 0.0354, "step": 8230 }, { - "epoch": 0.46, - "learning_rate": 1.0870120495307904e-05, - "loss": 0.0326, + "epoch": 0.91, + "learning_rate": 1.0869613505108842e-05, + "loss": 0.0868, "step": 8235 }, { - "epoch": 0.46, - "learning_rate": 1.086456771614193e-05, - "loss": 0.1464, + "epoch": 0.91, + "learning_rate": 1.0864060417592182e-05, + "loss": 0.1088, "step": 8240 }, { - "epoch": 0.46, - "learning_rate": 1.0859014936975958e-05, - "loss": 0.2912, + "epoch": 0.92, + "learning_rate": 1.0858507330075523e-05, + "loss": 0.1612, "step": 8245 }, { - "epoch": 0.46, - "learning_rate": 1.0853462157809985e-05, - "loss": 0.3831, + "epoch": 0.92, + "learning_rate": 1.0852954242558864e-05, + "loss": 0.235, "step": 8250 }, { - "epoch": 0.46, - "learning_rate": 1.0847909378644013e-05, - "loss": 0.0974, + "epoch": 0.92, + "learning_rate": 1.0847401155042205e-05, + "loss": 0.0823, "step": 8255 }, { - "epoch": 0.46, - "learning_rate": 1.0842356599478039e-05, - "loss": 0.1063, + "epoch": 0.92, + "learning_rate": 1.0841848067525544e-05, + "loss": 0.0721, "step": 8260 }, { - "epoch": 0.46, - "learning_rate": 1.0836803820312067e-05, - "loss": 0.1001, + "epoch": 0.92, + "learning_rate": 1.0836294980008885e-05, + "loss": 0.0969, "step": 8265 }, { - "epoch": 0.46, - "learning_rate": 1.0831251041146094e-05, - "loss": 0.0879, + "epoch": 0.92, + "learning_rate": 1.0830741892492226e-05, + "loss": 0.1035, "step": 8270 }, { - "epoch": 0.46, - "learning_rate": 1.0825698261980123e-05, - "loss": 0.1408, + "epoch": 0.92, + "learning_rate": 1.0825188804975567e-05, + "loss": 0.0786, "step": 8275 }, { - "epoch": 0.46, - "learning_rate": 1.0820145482814148e-05, - "loss": 0.0712, + "epoch": 0.92, + "learning_rate": 1.0819635717458907e-05, + "loss": 0.0562, "step": 8280 }, { - "epoch": 0.46, - "learning_rate": 1.0814592703648177e-05, - "loss": 0.0634, + "epoch": 0.92, + "learning_rate": 1.0814082629942248e-05, + "loss": 0.0721, "step": 8285 }, { - "epoch": 0.46, - "learning_rate": 1.0809039924482205e-05, - "loss": 0.1755, + "epoch": 0.92, + "learning_rate": 1.080852954242559e-05, + "loss": 0.1153, "step": 8290 }, { - "epoch": 0.46, - "learning_rate": 1.0803487145316232e-05, - "loss": 0.251, + "epoch": 0.92, + "learning_rate": 1.0802976454908932e-05, + "loss": 0.2193, "step": 8295 }, { - "epoch": 0.46, - "learning_rate": 1.079793436615026e-05, - "loss": 0.3524, + "epoch": 0.92, + "learning_rate": 1.079742336739227e-05, + "loss": 0.2758, "step": 8300 }, { - "epoch": 0.46, - "learning_rate": 1.0792381586984286e-05, - "loss": 0.0886, + "epoch": 0.92, + "learning_rate": 1.0791870279875612e-05, + "loss": 0.0604, "step": 8305 }, { - "epoch": 0.46, - "learning_rate": 1.0786828807818315e-05, - "loss": 0.1163, + "epoch": 0.92, + "learning_rate": 1.0786317192358953e-05, + "loss": 0.0817, "step": 8310 }, { - "epoch": 0.46, - "learning_rate": 1.078127602865234e-05, - "loss": 0.1324, + "epoch": 0.92, + "learning_rate": 1.0780764104842294e-05, + "loss": 0.0737, "step": 8315 }, { - "epoch": 0.46, - "learning_rate": 1.0775723249486369e-05, - "loss": 0.1298, + "epoch": 0.92, + "learning_rate": 1.0775211017325634e-05, + "loss": 0.0984, "step": 8320 }, { - "epoch": 0.46, - "learning_rate": 1.0770170470320396e-05, - "loss": 0.0973, + "epoch": 0.92, + "learning_rate": 1.0769657929808975e-05, + "loss": 0.0749, "step": 8325 }, { - "epoch": 0.46, - "learning_rate": 1.0764617691154424e-05, - "loss": 0.1354, + "epoch": 0.92, + "learning_rate": 1.0764104842292316e-05, + "loss": 0.0421, "step": 8330 }, { - "epoch": 0.46, - "learning_rate": 1.075906491198845e-05, - "loss": 0.0548, + "epoch": 0.93, + "learning_rate": 1.0758551754775657e-05, + "loss": 0.0684, "step": 8335 }, { - "epoch": 0.46, - "learning_rate": 1.0753512132822478e-05, - "loss": 0.1453, + "epoch": 0.93, + "learning_rate": 1.0752998667258996e-05, + "loss": 0.0969, "step": 8340 }, { - "epoch": 0.46, - "learning_rate": 1.0747959353656505e-05, - "loss": 0.2375, + "epoch": 0.93, + "learning_rate": 1.0747445579742337e-05, + "loss": 0.1404, "step": 8345 }, { - "epoch": 0.46, - "learning_rate": 1.0742406574490534e-05, - "loss": 0.3986, + "epoch": 0.93, + "learning_rate": 1.0741892492225678e-05, + "loss": 0.2184, "step": 8350 }, { - "epoch": 0.46, - "learning_rate": 1.073685379532456e-05, - "loss": 0.0823, + "epoch": 0.93, + "learning_rate": 1.073633940470902e-05, + "loss": 0.075, "step": 8355 }, { - "epoch": 0.46, - "learning_rate": 1.0731301016158588e-05, - "loss": 0.1077, + "epoch": 0.93, + "learning_rate": 1.0730786317192359e-05, + "loss": 0.083, "step": 8360 }, { - "epoch": 0.46, - "learning_rate": 1.0725748236992615e-05, - "loss": 0.1314, + "epoch": 0.93, + "learning_rate": 1.07252332296757e-05, + "loss": 0.0745, "step": 8365 }, { - "epoch": 0.46, - "learning_rate": 1.0720195457826643e-05, - "loss": 0.0978, + "epoch": 0.93, + "learning_rate": 1.071968014215904e-05, + "loss": 0.0828, "step": 8370 }, { - "epoch": 0.46, - "learning_rate": 1.0714642678660672e-05, - "loss": 0.1464, + "epoch": 0.93, + "learning_rate": 1.0714127054642382e-05, + "loss": 0.0698, "step": 8375 }, { - "epoch": 0.47, - "learning_rate": 1.0709089899494697e-05, - "loss": 0.0471, + "epoch": 0.93, + "learning_rate": 1.0708573967125725e-05, + "loss": 0.0825, "step": 8380 }, { - "epoch": 0.47, - "learning_rate": 1.0703537120328726e-05, - "loss": 0.1094, + "epoch": 0.93, + "learning_rate": 1.0703020879609062e-05, + "loss": 0.0779, "step": 8385 }, { - "epoch": 0.47, - "learning_rate": 1.0697984341162753e-05, - "loss": 0.1602, + "epoch": 0.93, + "learning_rate": 1.0697467792092403e-05, + "loss": 0.1404, "step": 8390 }, { - "epoch": 0.47, - "learning_rate": 1.0692431561996782e-05, - "loss": 0.1893, + "epoch": 0.93, + "learning_rate": 1.0691914704575746e-05, + "loss": 0.1043, "step": 8395 }, { - "epoch": 0.47, - "learning_rate": 1.0686878782830807e-05, - "loss": 0.3652, + "epoch": 0.93, + "learning_rate": 1.0686361617059087e-05, + "loss": 0.2751, "step": 8400 }, { - "epoch": 0.47, - "learning_rate": 1.0681326003664835e-05, - "loss": 0.1297, + "epoch": 0.93, + "learning_rate": 1.0680808529542427e-05, + "loss": 0.094, "step": 8405 }, { - "epoch": 0.47, - "learning_rate": 1.0675773224498862e-05, - "loss": 0.1068, + "epoch": 0.93, + "learning_rate": 1.0675255442025768e-05, + "loss": 0.0883, "step": 8410 }, { - "epoch": 0.47, - "learning_rate": 1.0670220445332891e-05, - "loss": 0.103, + "epoch": 0.93, + "learning_rate": 1.0669702354509109e-05, + "loss": 0.1024, "step": 8415 }, { - "epoch": 0.47, - "learning_rate": 1.0664667666166916e-05, - "loss": 0.1762, + "epoch": 0.93, + "learning_rate": 1.066414926699245e-05, + "loss": 0.0884, "step": 8420 }, { - "epoch": 0.47, - "learning_rate": 1.0659114887000945e-05, - "loss": 0.0928, + "epoch": 0.94, + "learning_rate": 1.0658596179475789e-05, + "loss": 0.0859, "step": 8425 }, { - "epoch": 0.47, - "learning_rate": 1.0653562107834972e-05, - "loss": 0.0699, + "epoch": 0.94, + "learning_rate": 1.065304309195913e-05, + "loss": 0.0794, "step": 8430 }, { - "epoch": 0.47, - "learning_rate": 1.0648009328669e-05, - "loss": 0.0709, + "epoch": 0.94, + "learning_rate": 1.0647490004442471e-05, + "loss": 0.0772, "step": 8435 }, { - "epoch": 0.47, - "learning_rate": 1.0642456549503026e-05, - "loss": 0.1662, + "epoch": 0.94, + "learning_rate": 1.0641936916925812e-05, + "loss": 0.0931, "step": 8440 }, { - "epoch": 0.47, - "learning_rate": 1.0636903770337054e-05, - "loss": 0.1954, + "epoch": 0.94, + "learning_rate": 1.0636383829409152e-05, + "loss": 0.1078, "step": 8445 }, { - "epoch": 0.47, - "learning_rate": 1.0631350991171081e-05, - "loss": 0.4755, + "epoch": 0.94, + "learning_rate": 1.0630830741892493e-05, + "loss": 0.2906, "step": 8450 }, { - "epoch": 0.47, - "learning_rate": 1.062579821200511e-05, - "loss": 0.0847, + "epoch": 0.94, + "learning_rate": 1.0625277654375834e-05, + "loss": 0.0769, "step": 8455 }, { - "epoch": 0.47, - "learning_rate": 1.0620245432839135e-05, - "loss": 0.1894, + "epoch": 0.94, + "learning_rate": 1.0619724566859175e-05, + "loss": 0.0995, "step": 8460 }, { - "epoch": 0.47, - "learning_rate": 1.0614692653673164e-05, - "loss": 0.1307, + "epoch": 0.94, + "learning_rate": 1.0614171479342514e-05, + "loss": 0.1122, "step": 8465 }, { - "epoch": 0.47, - "learning_rate": 1.0609139874507193e-05, - "loss": 0.1385, + "epoch": 0.94, + "learning_rate": 1.0608618391825855e-05, + "loss": 0.1084, "step": 8470 }, { - "epoch": 0.47, - "learning_rate": 1.060358709534122e-05, - "loss": 0.1344, + "epoch": 0.94, + "learning_rate": 1.0603065304309196e-05, + "loss": 0.0693, "step": 8475 }, { - "epoch": 0.47, - "learning_rate": 1.0598034316175248e-05, - "loss": 0.0474, + "epoch": 0.94, + "learning_rate": 1.0597512216792537e-05, + "loss": 0.0603, "step": 8480 }, { - "epoch": 0.47, - "learning_rate": 1.0592481537009273e-05, - "loss": 0.0685, + "epoch": 0.94, + "learning_rate": 1.0591959129275877e-05, + "loss": 0.0701, "step": 8485 }, { - "epoch": 0.47, - "learning_rate": 1.0586928757843302e-05, - "loss": 0.1659, + "epoch": 0.94, + "learning_rate": 1.0586406041759218e-05, + "loss": 0.1333, "step": 8490 }, { - "epoch": 0.47, - "learning_rate": 1.0581375978677329e-05, - "loss": 0.2401, + "epoch": 0.94, + "learning_rate": 1.058085295424256e-05, + "loss": 0.1673, "step": 8495 }, { - "epoch": 0.47, - "learning_rate": 1.0576933755344552e-05, - "loss": 0.3419, + "epoch": 0.94, + "learning_rate": 1.0575299866725902e-05, + "loss": 0.2957, "step": 8500 }, { - "epoch": 0.47, - "learning_rate": 1.0571380976178577e-05, - "loss": 0.1066, + "epoch": 0.94, + "learning_rate": 1.0569746779209243e-05, + "loss": 0.0753, "step": 8505 }, { - "epoch": 0.47, - "learning_rate": 1.0565828197012606e-05, - "loss": 0.1386, + "epoch": 0.94, + "learning_rate": 1.0564193691692582e-05, + "loss": 0.0823, "step": 8510 }, { - "epoch": 0.47, - "learning_rate": 1.0560275417846633e-05, - "loss": 0.101, + "epoch": 0.95, + "learning_rate": 1.0558640604175923e-05, + "loss": 0.081, "step": 8515 }, { - "epoch": 0.47, - "learning_rate": 1.0554722638680661e-05, - "loss": 0.1309, + "epoch": 0.95, + "learning_rate": 1.0553087516659264e-05, + "loss": 0.108, "step": 8520 }, { - "epoch": 0.47, - "learning_rate": 1.0549169859514687e-05, - "loss": 0.105, + "epoch": 0.95, + "learning_rate": 1.0547534429142605e-05, + "loss": 0.1037, "step": 8525 }, { - "epoch": 0.47, - "learning_rate": 1.0543617080348715e-05, - "loss": 0.099, + "epoch": 0.95, + "learning_rate": 1.0541981341625945e-05, + "loss": 0.0428, "step": 8530 }, { - "epoch": 0.47, - "learning_rate": 1.0538064301182742e-05, - "loss": 0.1338, + "epoch": 0.95, + "learning_rate": 1.0536428254109286e-05, + "loss": 0.0715, "step": 8535 }, { - "epoch": 0.47, - "learning_rate": 1.0532511522016771e-05, - "loss": 0.1342, + "epoch": 0.95, + "learning_rate": 1.0530875166592627e-05, + "loss": 0.0539, "step": 8540 }, { - "epoch": 0.47, - "learning_rate": 1.0526958742850796e-05, - "loss": 0.1997, + "epoch": 0.95, + "learning_rate": 1.0525322079075968e-05, + "loss": 0.1419, "step": 8545 }, { - "epoch": 0.47, - "learning_rate": 1.0521405963684825e-05, - "loss": 0.45, + "epoch": 0.95, + "learning_rate": 1.0519768991559307e-05, + "loss": 0.2654, "step": 8550 }, { - "epoch": 0.47, - "learning_rate": 1.0515853184518852e-05, - "loss": 0.0777, + "epoch": 0.95, + "learning_rate": 1.0514215904042648e-05, + "loss": 0.0642, "step": 8555 }, { - "epoch": 0.48, - "learning_rate": 1.051030040535288e-05, - "loss": 0.1159, + "epoch": 0.95, + "learning_rate": 1.050866281652599e-05, + "loss": 0.0741, "step": 8560 }, { - "epoch": 0.48, - "learning_rate": 1.0504747626186906e-05, - "loss": 0.1564, + "epoch": 0.95, + "learning_rate": 1.050310972900933e-05, + "loss": 0.0991, "step": 8565 }, { - "epoch": 0.48, - "learning_rate": 1.0499194847020934e-05, - "loss": 0.1661, + "epoch": 0.95, + "learning_rate": 1.049755664149267e-05, + "loss": 0.0959, "step": 8570 }, { - "epoch": 0.48, - "learning_rate": 1.0493642067854963e-05, - "loss": 0.0812, + "epoch": 0.95, + "learning_rate": 1.049200355397601e-05, + "loss": 0.0606, "step": 8575 }, { - "epoch": 0.48, - "learning_rate": 1.048808928868899e-05, - "loss": 0.0601, + "epoch": 0.95, + "learning_rate": 1.0486450466459352e-05, + "loss": 0.0777, "step": 8580 }, { - "epoch": 0.48, - "learning_rate": 1.0482536509523018e-05, - "loss": 0.0998, + "epoch": 0.95, + "learning_rate": 1.0480897378942693e-05, + "loss": 0.0993, "step": 8585 }, { - "epoch": 0.48, - "learning_rate": 1.0476983730357044e-05, - "loss": 0.0926, + "epoch": 0.95, + "learning_rate": 1.0475344291426032e-05, + "loss": 0.0984, "step": 8590 }, { - "epoch": 0.48, - "learning_rate": 1.0471430951191072e-05, - "loss": 0.2665, + "epoch": 0.95, + "learning_rate": 1.0469791203909373e-05, + "loss": 0.1397, "step": 8595 }, { - "epoch": 0.48, - "learning_rate": 1.04658781720251e-05, - "loss": 0.4059, + "epoch": 0.95, + "learning_rate": 1.0464238116392716e-05, + "loss": 0.2935, "step": 8600 }, { - "epoch": 0.48, - "learning_rate": 1.0460325392859128e-05, - "loss": 0.139, + "epoch": 0.96, + "learning_rate": 1.0458685028876057e-05, + "loss": 0.0781, "step": 8605 }, { - "epoch": 0.48, - "learning_rate": 1.0454772613693153e-05, - "loss": 0.1159, + "epoch": 0.96, + "learning_rate": 1.0453131941359395e-05, + "loss": 0.0568, "step": 8610 }, { - "epoch": 0.48, - "learning_rate": 1.0449219834527182e-05, - "loss": 0.1026, + "epoch": 0.96, + "learning_rate": 1.0447578853842738e-05, + "loss": 0.1006, "step": 8615 }, { - "epoch": 0.48, - "learning_rate": 1.0443667055361209e-05, - "loss": 0.1513, + "epoch": 0.96, + "learning_rate": 1.0442025766326079e-05, + "loss": 0.0871, "step": 8620 }, { - "epoch": 0.48, - "learning_rate": 1.0438114276195237e-05, - "loss": 0.1182, + "epoch": 0.96, + "learning_rate": 1.043647267880942e-05, + "loss": 0.0733, "step": 8625 }, { - "epoch": 0.48, - "learning_rate": 1.0432561497029263e-05, - "loss": 0.0361, + "epoch": 0.96, + "learning_rate": 1.0430919591292759e-05, + "loss": 0.0508, "step": 8630 }, { - "epoch": 0.48, - "learning_rate": 1.0427008717863291e-05, - "loss": 0.0828, + "epoch": 0.96, + "learning_rate": 1.04253665037761e-05, + "loss": 0.0861, "step": 8635 }, { - "epoch": 0.48, - "learning_rate": 1.0421455938697318e-05, - "loss": 0.1605, + "epoch": 0.96, + "learning_rate": 1.0419813416259441e-05, + "loss": 0.1221, "step": 8640 }, { - "epoch": 0.48, - "learning_rate": 1.0415903159531347e-05, - "loss": 0.156, + "epoch": 0.96, + "learning_rate": 1.0414260328742782e-05, + "loss": 0.1375, "step": 8645 }, { - "epoch": 0.48, - "learning_rate": 1.0410350380365372e-05, - "loss": 0.2625, + "epoch": 0.96, + "learning_rate": 1.0408707241226123e-05, + "loss": 0.2757, "step": 8650 }, { - "epoch": 0.48, - "learning_rate": 1.0404797601199401e-05, - "loss": 0.1392, + "epoch": 0.96, + "learning_rate": 1.0403154153709463e-05, + "loss": 0.072, "step": 8655 }, { - "epoch": 0.48, - "learning_rate": 1.039924482203343e-05, - "loss": 0.0905, + "epoch": 0.96, + "learning_rate": 1.0397601066192804e-05, + "loss": 0.0904, "step": 8660 }, { - "epoch": 0.48, - "learning_rate": 1.0393692042867456e-05, - "loss": 0.139, + "epoch": 0.96, + "learning_rate": 1.0392047978676145e-05, + "loss": 0.0962, "step": 8665 }, { - "epoch": 0.48, - "learning_rate": 1.0388139263701485e-05, - "loss": 0.0935, + "epoch": 0.96, + "learning_rate": 1.0386494891159486e-05, + "loss": 0.054, "step": 8670 }, { - "epoch": 0.48, - "learning_rate": 1.038258648453551e-05, - "loss": 0.1118, + "epoch": 0.96, + "learning_rate": 1.0380941803642825e-05, + "loss": 0.0826, "step": 8675 }, { - "epoch": 0.48, - "learning_rate": 1.0377033705369539e-05, - "loss": 0.0338, + "epoch": 0.96, + "learning_rate": 1.0375388716126166e-05, + "loss": 0.0445, "step": 8680 }, { - "epoch": 0.48, - "learning_rate": 1.0371480926203566e-05, - "loss": 0.0682, + "epoch": 0.96, + "learning_rate": 1.0369835628609507e-05, + "loss": 0.0709, "step": 8685 }, { - "epoch": 0.48, - "learning_rate": 1.0365928147037595e-05, - "loss": 0.1566, + "epoch": 0.96, + "learning_rate": 1.036428254109285e-05, + "loss": 0.0767, "step": 8690 }, { - "epoch": 0.48, - "learning_rate": 1.036037536787162e-05, - "loss": 0.1398, + "epoch": 0.97, + "learning_rate": 1.0358729453576188e-05, + "loss": 0.1863, "step": 8695 }, { - "epoch": 0.48, - "learning_rate": 1.0354822588705648e-05, - "loss": 0.3807, + "epoch": 0.97, + "learning_rate": 1.0353176366059529e-05, + "loss": 0.3167, "step": 8700 }, { - "epoch": 0.48, - "learning_rate": 1.0349269809539675e-05, - "loss": 0.1135, + "epoch": 0.97, + "learning_rate": 1.0347623278542872e-05, + "loss": 0.0604, "step": 8705 }, { - "epoch": 0.48, - "learning_rate": 1.0343717030373704e-05, - "loss": 0.0983, + "epoch": 0.97, + "learning_rate": 1.0342070191026213e-05, + "loss": 0.0765, "step": 8710 }, { - "epoch": 0.48, - "learning_rate": 1.033816425120773e-05, - "loss": 0.1753, + "epoch": 0.97, + "learning_rate": 1.0336517103509552e-05, + "loss": 0.0985, "step": 8715 }, { - "epoch": 0.48, - "learning_rate": 1.0332611472041758e-05, - "loss": 0.1595, + "epoch": 0.97, + "learning_rate": 1.0330964015992893e-05, + "loss": 0.1063, "step": 8720 }, { - "epoch": 0.48, - "learning_rate": 1.0327058692875785e-05, - "loss": 0.0795, + "epoch": 0.97, + "learning_rate": 1.0325410928476234e-05, + "loss": 0.0517, "step": 8725 }, { - "epoch": 0.48, - "learning_rate": 1.0321505913709814e-05, - "loss": 0.0854, + "epoch": 0.97, + "learning_rate": 1.0319857840959575e-05, + "loss": 0.0216, "step": 8730 }, { - "epoch": 0.48, - "learning_rate": 1.0315953134543839e-05, - "loss": 0.1353, + "epoch": 0.97, + "learning_rate": 1.0314304753442915e-05, + "loss": 0.0608, "step": 8735 }, { - "epoch": 0.49, - "learning_rate": 1.0310400355377867e-05, - "loss": 0.1933, + "epoch": 0.97, + "learning_rate": 1.0308751665926256e-05, + "loss": 0.1446, "step": 8740 }, { - "epoch": 0.49, - "learning_rate": 1.0304847576211894e-05, - "loss": 0.1334, + "epoch": 0.97, + "learning_rate": 1.0303198578409597e-05, + "loss": 0.1741, "step": 8745 }, { - "epoch": 0.49, - "learning_rate": 1.0299294797045923e-05, - "loss": 0.4577, + "epoch": 0.97, + "learning_rate": 1.0297645490892938e-05, + "loss": 0.193, "step": 8750 }, { - "epoch": 0.49, - "learning_rate": 1.029374201787995e-05, - "loss": 0.1197, + "epoch": 0.97, + "learning_rate": 1.0292092403376277e-05, + "loss": 0.0618, "step": 8755 }, { - "epoch": 0.49, - "learning_rate": 1.0288189238713977e-05, - "loss": 0.1421, + "epoch": 0.97, + "learning_rate": 1.0286539315859618e-05, + "loss": 0.0628, "step": 8760 }, { - "epoch": 0.49, - "learning_rate": 1.0282636459548006e-05, - "loss": 0.1662, + "epoch": 0.97, + "learning_rate": 1.028098622834296e-05, + "loss": 0.1094, "step": 8765 }, { - "epoch": 0.49, - "learning_rate": 1.0277083680382033e-05, - "loss": 0.1282, + "epoch": 0.97, + "learning_rate": 1.02754331408263e-05, + "loss": 0.0873, "step": 8770 }, { - "epoch": 0.49, - "learning_rate": 1.027153090121606e-05, - "loss": 0.0732, + "epoch": 0.97, + "learning_rate": 1.0269880053309641e-05, + "loss": 0.0766, "step": 8775 }, { - "epoch": 0.49, - "learning_rate": 1.0265978122050086e-05, - "loss": 0.1477, + "epoch": 0.97, + "learning_rate": 1.026432696579298e-05, + "loss": 0.0489, "step": 8780 }, { - "epoch": 0.49, - "learning_rate": 1.0260425342884115e-05, - "loss": 0.1379, + "epoch": 0.98, + "learning_rate": 1.0258773878276322e-05, + "loss": 0.0592, "step": 8785 }, { - "epoch": 0.49, - "learning_rate": 1.0254872563718142e-05, - "loss": 0.1464, + "epoch": 0.98, + "learning_rate": 1.0253220790759663e-05, + "loss": 0.0832, "step": 8790 }, { - "epoch": 0.49, - "learning_rate": 1.0249319784552169e-05, - "loss": 0.2031, + "epoch": 0.98, + "learning_rate": 1.0247667703243006e-05, + "loss": 0.1182, "step": 8795 }, { - "epoch": 0.49, - "learning_rate": 1.0243767005386196e-05, - "loss": 0.3198, + "epoch": 0.98, + "learning_rate": 1.0242114615726343e-05, + "loss": 0.3355, "step": 8800 }, { - "epoch": 0.49, - "learning_rate": 1.0238214226220225e-05, - "loss": 0.0853, + "epoch": 0.98, + "learning_rate": 1.0236561528209686e-05, + "loss": 0.0623, "step": 8805 }, { - "epoch": 0.49, - "learning_rate": 1.0232661447054252e-05, - "loss": 0.0932, + "epoch": 0.98, + "learning_rate": 1.0231008440693027e-05, + "loss": 0.0657, "step": 8810 }, { - "epoch": 0.49, - "learning_rate": 1.0227108667888278e-05, - "loss": 0.0997, + "epoch": 0.98, + "learning_rate": 1.0225455353176368e-05, + "loss": 0.0858, "step": 8815 }, { - "epoch": 0.49, - "learning_rate": 1.0221555888722305e-05, - "loss": 0.0998, + "epoch": 0.98, + "learning_rate": 1.0219902265659708e-05, + "loss": 0.0962, "step": 8820 }, { - "epoch": 0.49, - "learning_rate": 1.0216003109556334e-05, - "loss": 0.0792, + "epoch": 0.98, + "learning_rate": 1.0214349178143049e-05, + "loss": 0.0589, "step": 8825 }, { - "epoch": 0.49, - "learning_rate": 1.0210450330390361e-05, - "loss": 0.1122, + "epoch": 0.98, + "learning_rate": 1.020879609062639e-05, + "loss": 0.1494, "step": 8830 }, { - "epoch": 0.49, - "learning_rate": 1.0204897551224388e-05, - "loss": 0.0986, + "epoch": 0.98, + "learning_rate": 1.020324300310973e-05, + "loss": 0.0902, "step": 8835 }, { - "epoch": 0.49, - "learning_rate": 1.0199344772058417e-05, - "loss": 0.0995, + "epoch": 0.98, + "learning_rate": 1.019768991559307e-05, + "loss": 0.1051, "step": 8840 }, { - "epoch": 0.49, - "learning_rate": 1.0193791992892444e-05, - "loss": 0.251, + "epoch": 0.98, + "learning_rate": 1.0192136828076411e-05, + "loss": 0.1033, "step": 8845 }, { - "epoch": 0.49, - "learning_rate": 1.0188239213726472e-05, - "loss": 0.3958, + "epoch": 0.98, + "learning_rate": 1.0186583740559752e-05, + "loss": 0.3234, "step": 8850 }, { - "epoch": 0.49, - "learning_rate": 1.0182686434560497e-05, - "loss": 0.1468, + "epoch": 0.98, + "learning_rate": 1.0181030653043093e-05, + "loss": 0.0701, "step": 8855 }, { - "epoch": 0.49, - "learning_rate": 1.0177133655394526e-05, - "loss": 0.0893, + "epoch": 0.98, + "learning_rate": 1.0175477565526433e-05, + "loss": 0.0662, "step": 8860 }, { - "epoch": 0.49, - "learning_rate": 1.0171580876228553e-05, - "loss": 0.12, + "epoch": 0.98, + "learning_rate": 1.0169924478009774e-05, + "loss": 0.0727, "step": 8865 }, { - "epoch": 0.49, - "learning_rate": 1.0166028097062582e-05, - "loss": 0.1379, + "epoch": 0.98, + "learning_rate": 1.0164371390493115e-05, + "loss": 0.1058, "step": 8870 }, { - "epoch": 0.49, - "learning_rate": 1.0160475317896607e-05, - "loss": 0.1327, + "epoch": 0.99, + "learning_rate": 1.0158818302976456e-05, + "loss": 0.0291, "step": 8875 }, { - "epoch": 0.49, - "learning_rate": 1.0154922538730636e-05, - "loss": 0.0665, + "epoch": 0.99, + "learning_rate": 1.0153265215459795e-05, + "loss": 0.0796, "step": 8880 }, { - "epoch": 0.49, - "learning_rate": 1.0149369759564663e-05, - "loss": 0.1, + "epoch": 0.99, + "learning_rate": 1.0147712127943136e-05, + "loss": 0.1045, "step": 8885 }, { - "epoch": 0.49, - "learning_rate": 1.0143816980398691e-05, - "loss": 0.1341, + "epoch": 0.99, + "learning_rate": 1.0142159040426477e-05, + "loss": 0.1044, "step": 8890 }, { - "epoch": 0.49, - "learning_rate": 1.0138264201232716e-05, - "loss": 0.264, + "epoch": 0.99, + "learning_rate": 1.0136605952909818e-05, + "loss": 0.189, "step": 8895 }, { - "epoch": 0.49, - "learning_rate": 1.0132711422066745e-05, - "loss": 0.5675, + "epoch": 0.99, + "learning_rate": 1.0131052865393158e-05, + "loss": 0.2141, "step": 8900 }, { - "epoch": 0.49, - "learning_rate": 1.0127158642900772e-05, - "loss": 0.1239, + "epoch": 0.99, + "learning_rate": 1.0125499777876499e-05, + "loss": 0.0881, "step": 8905 }, { - "epoch": 0.49, - "learning_rate": 1.01216058637348e-05, - "loss": 0.1045, + "epoch": 0.99, + "learning_rate": 1.0119946690359842e-05, + "loss": 0.0875, "step": 8910 }, { - "epoch": 0.49, - "learning_rate": 1.0116053084568826e-05, - "loss": 0.0927, + "epoch": 0.99, + "learning_rate": 1.0114393602843183e-05, + "loss": 0.0863, "step": 8915 }, { - "epoch": 0.5, - "learning_rate": 1.0110500305402855e-05, - "loss": 0.1284, + "epoch": 0.99, + "learning_rate": 1.0108840515326524e-05, + "loss": 0.089, "step": 8920 }, { - "epoch": 0.5, - "learning_rate": 1.0104947526236883e-05, - "loss": 0.0436, + "epoch": 0.99, + "learning_rate": 1.0103287427809863e-05, + "loss": 0.0462, "step": 8925 }, { - "epoch": 0.5, - "learning_rate": 1.009939474707091e-05, - "loss": 0.071, + "epoch": 0.99, + "learning_rate": 1.0097734340293204e-05, + "loss": 0.0526, "step": 8930 }, { - "epoch": 0.5, - "learning_rate": 1.0093841967904939e-05, - "loss": 0.1141, + "epoch": 0.99, + "learning_rate": 1.0092181252776545e-05, + "loss": 0.0568, "step": 8935 }, { - "epoch": 0.5, - "learning_rate": 1.0088289188738964e-05, - "loss": 0.1236, + "epoch": 0.99, + "learning_rate": 1.0086628165259886e-05, + "loss": 0.1226, "step": 8940 }, { - "epoch": 0.5, - "learning_rate": 1.0082736409572993e-05, - "loss": 0.2995, + "epoch": 0.99, + "learning_rate": 1.0081075077743226e-05, + "loss": 0.2144, "step": 8945 }, { - "epoch": 0.5, - "learning_rate": 1.007718363040702e-05, - "loss": 0.3582, + "epoch": 0.99, + "learning_rate": 1.0075521990226567e-05, + "loss": 0.3195, "step": 8950 }, { - "epoch": 0.5, - "learning_rate": 1.0071630851241048e-05, - "loss": 0.1228, + "epoch": 0.99, + "learning_rate": 1.0069968902709908e-05, + "loss": 0.0534, "step": 8955 }, { - "epoch": 0.5, - "learning_rate": 1.0066078072075074e-05, - "loss": 0.1174, + "epoch": 0.99, + "learning_rate": 1.0064415815193249e-05, + "loss": 0.0817, "step": 8960 }, { - "epoch": 0.5, - "learning_rate": 1.0060525292909102e-05, - "loss": 0.1173, + "epoch": 1.0, + "learning_rate": 1.0058862727676588e-05, + "loss": 0.0665, "step": 8965 }, { - "epoch": 0.5, - "learning_rate": 1.0054972513743129e-05, - "loss": 0.1144, + "epoch": 1.0, + "learning_rate": 1.005330964015993e-05, + "loss": 0.0908, "step": 8970 }, { - "epoch": 0.5, - "learning_rate": 1.0049419734577158e-05, - "loss": 0.1437, + "epoch": 1.0, + "learning_rate": 1.004775655264327e-05, + "loss": 0.0749, "step": 8975 }, { - "epoch": 0.5, - "learning_rate": 1.0043866955411183e-05, - "loss": 0.0374, + "epoch": 1.0, + "learning_rate": 1.0042203465126611e-05, + "loss": 0.044, "step": 8980 }, { - "epoch": 0.5, - "learning_rate": 1.0038314176245212e-05, - "loss": 0.0984, + "epoch": 1.0, + "learning_rate": 1.003665037760995e-05, + "loss": 0.0613, "step": 8985 }, { - "epoch": 0.5, - "learning_rate": 1.0032761397079239e-05, - "loss": 0.0884, + "epoch": 1.0, + "learning_rate": 1.0031097290093292e-05, + "loss": 0.1023, "step": 8990 }, { - "epoch": 0.5, - "learning_rate": 1.0027208617913267e-05, - "loss": 0.1086, + "epoch": 1.0, + "learning_rate": 1.0025544202576633e-05, + "loss": 0.143, "step": 8995 }, { - "epoch": 0.5, - "learning_rate": 1.0021655838747292e-05, - "loss": 0.3245, + "epoch": 1.0, + "learning_rate": 1.0019991115059976e-05, + "loss": 0.2331, "step": 9000 }, { - "epoch": 0.5, - "learning_rate": 1.0016103059581321e-05, - "loss": 0.1096, + "epoch": 1.0, + "learning_rate": 1.0014438027543313e-05, + "loss": 0.0839, "step": 9005 }, { - "epoch": 0.5, - "learning_rate": 1.0010550280415348e-05, - "loss": 0.099, + "epoch": 1.0, + "learning_rate": 1.0008884940026654e-05, + "loss": 0.1403, "step": 9010 }, { - "epoch": 0.5, - "learning_rate": 1.0004997501249377e-05, - "loss": 0.125, + "epoch": 1.0, + "learning_rate": 1.0003331852509997e-05, + "loss": 0.0757, "step": 9015 }, { - "epoch": 0.5, - "learning_rate": 9.999444722083404e-06, - "loss": 0.1397, + "epoch": 1.0, + "learning_rate": 9.997778764993337e-06, + "loss": 0.093, "step": 9020 }, { - "epoch": 0.5, - "learning_rate": 9.99389194291743e-06, - "loss": 0.1217, + "epoch": 1.0, + "learning_rate": 9.992225677476678e-06, + "loss": 0.0854, "step": 9025 }, { - "epoch": 0.5, - "learning_rate": 9.988339163751458e-06, - "loss": 0.0598, + "epoch": 1.0, + "learning_rate": 9.986672589960019e-06, + "loss": 0.084, "step": 9030 }, { - "epoch": 0.5, - "learning_rate": 9.982786384585486e-06, - "loss": 0.0882, + "epoch": 1.0, + "learning_rate": 9.98111950244336e-06, + "loss": 0.0606, "step": 9035 }, { - "epoch": 0.5, - "learning_rate": 9.977233605419513e-06, - "loss": 0.2148, + "epoch": 1.0, + "learning_rate": 9.975566414926699e-06, + "loss": 0.051, "step": 9040 }, { - "epoch": 0.5, - "learning_rate": 9.97168082625354e-06, - "loss": 0.1438, + "epoch": 1.0, + "learning_rate": 9.97001332741004e-06, + "loss": 0.0723, "step": 9045 }, { - "epoch": 0.5, - "learning_rate": 9.966128047087567e-06, - "loss": 0.3344, + "epoch": 1.0, + "learning_rate": 9.964460239893381e-06, + "loss": 0.1134, "step": 9050 }, { - "epoch": 0.5, - "learning_rate": 9.960575267921596e-06, - "loss": 0.0786, + "epoch": 1.01, + "learning_rate": 9.958907152376722e-06, + "loss": 0.2029, "step": 9055 }, { - "epoch": 0.5, - "learning_rate": 9.955022488755623e-06, - "loss": 0.0914, + "epoch": 1.01, + "learning_rate": 9.953354064860063e-06, + "loss": 0.2837, "step": 9060 }, { - "epoch": 0.5, - "learning_rate": 9.949469709589651e-06, - "loss": 0.1258, + "epoch": 1.01, + "learning_rate": 9.947800977343404e-06, + "loss": 0.0507, "step": 9065 }, { - "epoch": 0.5, - "learning_rate": 9.943916930423678e-06, - "loss": 0.1213, + "epoch": 1.01, + "learning_rate": 9.942247889826745e-06, + "loss": 0.0734, "step": 9070 }, { - "epoch": 0.5, - "learning_rate": 9.938364151257705e-06, - "loss": 0.131, + "epoch": 1.01, + "learning_rate": 9.936694802310085e-06, + "loss": 0.0853, "step": 9075 }, { - "epoch": 0.5, - "learning_rate": 9.932811372091732e-06, - "loss": 0.0818, + "epoch": 1.01, + "learning_rate": 9.931141714793426e-06, + "loss": 0.0952, "step": 9080 }, { - "epoch": 0.5, - "learning_rate": 9.92725859292576e-06, - "loss": 0.1002, + "epoch": 1.01, + "learning_rate": 9.925588627276767e-06, + "loss": 0.0532, "step": 9085 }, { - "epoch": 0.5, - "learning_rate": 9.921705813759788e-06, - "loss": 0.1833, + "epoch": 1.01, + "learning_rate": 9.920035539760108e-06, + "loss": 0.0337, "step": 9090 }, { - "epoch": 0.5, - "learning_rate": 9.916153034593815e-06, - "loss": 0.1473, + "epoch": 1.01, + "learning_rate": 9.914482452243447e-06, + "loss": 0.0653, "step": 9095 }, { - "epoch": 0.51, - "learning_rate": 9.910600255427842e-06, - "loss": 0.2721, + "epoch": 1.01, + "learning_rate": 9.908929364726788e-06, + "loss": 0.0994, "step": 9100 }, { - "epoch": 0.51, - "learning_rate": 9.90504747626187e-06, - "loss": 0.1278, + "epoch": 1.01, + "learning_rate": 9.90337627721013e-06, + "loss": 0.0857, "step": 9105 }, { - "epoch": 0.51, - "learning_rate": 9.899494697095897e-06, - "loss": 0.1003, + "epoch": 1.01, + "learning_rate": 9.89782318969347e-06, + "loss": 0.2042, "step": 9110 }, { - "epoch": 0.51, - "learning_rate": 9.893941917929924e-06, - "loss": 0.131, + "epoch": 1.01, + "learning_rate": 9.892270102176812e-06, + "loss": 0.07, "step": 9115 }, { - "epoch": 0.51, - "learning_rate": 9.888389138763951e-06, - "loss": 0.1129, + "epoch": 1.01, + "learning_rate": 9.886717014660153e-06, + "loss": 0.0933, "step": 9120 }, { - "epoch": 0.51, - "learning_rate": 9.88283635959798e-06, - "loss": 0.0835, + "epoch": 1.01, + "learning_rate": 9.881163927143492e-06, + "loss": 0.063, "step": 9125 }, { - "epoch": 0.51, - "learning_rate": 9.877283580432007e-06, - "loss": 0.0487, + "epoch": 1.01, + "learning_rate": 9.875610839626833e-06, + "loss": 0.0854, "step": 9130 }, { - "epoch": 0.51, - "learning_rate": 9.871730801266034e-06, - "loss": 0.0866, + "epoch": 1.01, + "learning_rate": 9.870057752110174e-06, + "loss": 0.0575, "step": 9135 }, { - "epoch": 0.51, - "learning_rate": 9.86617802210006e-06, - "loss": 0.0777, + "epoch": 1.01, + "learning_rate": 9.864504664593515e-06, + "loss": 0.0519, "step": 9140 }, { - "epoch": 0.51, - "learning_rate": 9.86062524293409e-06, - "loss": 0.2873, + "epoch": 1.02, + "learning_rate": 9.858951577076855e-06, + "loss": 0.0573, "step": 9145 }, { - "epoch": 0.51, - "learning_rate": 9.855072463768118e-06, - "loss": 0.4195, + "epoch": 1.02, + "learning_rate": 9.853398489560196e-06, + "loss": 0.1191, "step": 9150 }, { - "epoch": 0.51, - "learning_rate": 9.849519684602145e-06, - "loss": 0.0778, + "epoch": 1.02, + "learning_rate": 9.847845402043537e-06, + "loss": 0.1459, "step": 9155 }, { - "epoch": 0.51, - "learning_rate": 9.843966905436172e-06, - "loss": 0.1094, + "epoch": 1.02, + "learning_rate": 9.842292314526878e-06, + "loss": 0.2295, "step": 9160 }, { - "epoch": 0.51, - "learning_rate": 9.838414126270199e-06, - "loss": 0.131, + "epoch": 1.02, + "learning_rate": 9.836739227010217e-06, + "loss": 0.0642, "step": 9165 }, { - "epoch": 0.51, - "learning_rate": 9.832861347104227e-06, - "loss": 0.1448, + "epoch": 1.02, + "learning_rate": 9.83118613949356e-06, + "loss": 0.0675, "step": 9170 }, { - "epoch": 0.51, - "learning_rate": 9.827308567938254e-06, - "loss": 0.0768, + "epoch": 1.02, + "learning_rate": 9.8256330519769e-06, + "loss": 0.0686, "step": 9175 }, { - "epoch": 0.51, - "learning_rate": 9.821755788772281e-06, - "loss": 0.1119, + "epoch": 1.02, + "learning_rate": 9.82007996446024e-06, + "loss": 0.1142, "step": 9180 }, { - "epoch": 0.51, - "learning_rate": 9.816203009606308e-06, - "loss": 0.1849, + "epoch": 1.02, + "learning_rate": 9.814526876943581e-06, + "loss": 0.0696, "step": 9185 }, { - "epoch": 0.51, - "learning_rate": 9.810650230440337e-06, - "loss": 0.2016, + "epoch": 1.02, + "learning_rate": 9.808973789426923e-06, + "loss": 0.0416, "step": 9190 }, { - "epoch": 0.51, - "learning_rate": 9.805097451274364e-06, - "loss": 0.159, + "epoch": 1.02, + "learning_rate": 9.803420701910264e-06, + "loss": 0.0972, "step": 9195 }, { - "epoch": 0.51, - "learning_rate": 9.79954467210839e-06, - "loss": 0.3989, + "epoch": 1.02, + "learning_rate": 9.797867614393603e-06, + "loss": 0.1283, "step": 9200 }, { - "epoch": 0.51, - "learning_rate": 9.793991892942418e-06, - "loss": 0.0607, + "epoch": 1.02, + "learning_rate": 9.792314526876944e-06, + "loss": 0.1552, "step": 9205 }, { - "epoch": 0.51, - "learning_rate": 9.788439113776446e-06, - "loss": 0.0891, + "epoch": 1.02, + "learning_rate": 9.786761439360285e-06, + "loss": 0.2492, "step": 9210 }, { - "epoch": 0.51, - "learning_rate": 9.782886334610473e-06, - "loss": 0.1278, + "epoch": 1.02, + "learning_rate": 9.781208351843626e-06, + "loss": 0.0781, "step": 9215 }, { - "epoch": 0.51, - "learning_rate": 9.7773335554445e-06, - "loss": 0.1269, + "epoch": 1.02, + "learning_rate": 9.775655264326967e-06, + "loss": 0.0934, "step": 9220 }, { - "epoch": 0.51, - "learning_rate": 9.771780776278527e-06, - "loss": 0.1155, + "epoch": 1.02, + "learning_rate": 9.770102176810308e-06, + "loss": 0.0825, "step": 9225 }, { - "epoch": 0.51, - "learning_rate": 9.766227997112556e-06, - "loss": 0.0392, + "epoch": 1.02, + "learning_rate": 9.764549089293648e-06, + "loss": 0.1004, "step": 9230 }, { - "epoch": 0.51, - "learning_rate": 9.760675217946583e-06, - "loss": 0.0887, + "epoch": 1.03, + "learning_rate": 9.758996001776989e-06, + "loss": 0.0455, "step": 9235 }, { - "epoch": 0.51, - "learning_rate": 9.755122438780611e-06, - "loss": 0.0851, + "epoch": 1.03, + "learning_rate": 9.75344291426033e-06, + "loss": 0.084, "step": 9240 }, { - "epoch": 0.51, - "learning_rate": 9.749569659614638e-06, - "loss": 0.1781, + "epoch": 1.03, + "learning_rate": 9.74788982674367e-06, + "loss": 0.0802, "step": 9245 }, { - "epoch": 0.51, - "learning_rate": 9.744016880448665e-06, - "loss": 0.2902, + "epoch": 1.03, + "learning_rate": 9.74233673922701e-06, + "loss": 0.1308, "step": 9250 }, { - "epoch": 0.51, - "learning_rate": 9.738464101282692e-06, - "loss": 0.1201, + "epoch": 1.03, + "learning_rate": 9.736783651710351e-06, + "loss": 0.1009, "step": 9255 }, { - "epoch": 0.51, - "learning_rate": 9.732911322116721e-06, - "loss": 0.1322, + "epoch": 1.03, + "learning_rate": 9.731230564193692e-06, + "loss": 0.2178, "step": 9260 }, { - "epoch": 0.51, - "learning_rate": 9.727358542950748e-06, - "loss": 0.1361, + "epoch": 1.03, + "learning_rate": 9.725677476677033e-06, + "loss": 0.0614, "step": 9265 }, { - "epoch": 0.51, - "learning_rate": 9.721805763784775e-06, - "loss": 0.1449, + "epoch": 1.03, + "learning_rate": 9.720124389160374e-06, + "loss": 0.1016, "step": 9270 }, { - "epoch": 0.51, - "learning_rate": 9.716252984618802e-06, - "loss": 0.091, + "epoch": 1.03, + "learning_rate": 9.714571301643715e-06, + "loss": 0.0839, "step": 9275 }, { - "epoch": 0.51, - "learning_rate": 9.71070020545283e-06, - "loss": 0.0693, + "epoch": 1.03, + "learning_rate": 9.709018214127055e-06, + "loss": 0.0835, "step": 9280 }, { - "epoch": 0.52, - "learning_rate": 9.705147426286857e-06, - "loss": 0.0691, + "epoch": 1.03, + "learning_rate": 9.703465126610396e-06, + "loss": 0.0296, "step": 9285 }, { - "epoch": 0.52, - "learning_rate": 9.699594647120884e-06, - "loss": 0.1784, + "epoch": 1.03, + "learning_rate": 9.697912039093737e-06, + "loss": 0.0332, "step": 9290 }, { - "epoch": 0.52, - "learning_rate": 9.694041867954911e-06, - "loss": 0.2105, + "epoch": 1.03, + "learning_rate": 9.692358951577078e-06, + "loss": 0.0958, "step": 9295 }, { - "epoch": 0.52, - "learning_rate": 9.68848908878894e-06, - "loss": 0.2459, + "epoch": 1.03, + "learning_rate": 9.686805864060417e-06, + "loss": 0.1346, "step": 9300 }, { - "epoch": 0.52, - "learning_rate": 9.682936309622967e-06, - "loss": 0.141, + "epoch": 1.03, + "learning_rate": 9.681252776543758e-06, + "loss": 0.1797, "step": 9305 }, { - "epoch": 0.52, - "learning_rate": 9.677383530456994e-06, - "loss": 0.1422, + "epoch": 1.03, + "learning_rate": 9.6756996890271e-06, + "loss": 0.2794, "step": 9310 }, { - "epoch": 0.52, - "learning_rate": 9.67183075129102e-06, - "loss": 0.1236, + "epoch": 1.03, + "learning_rate": 9.67014660151044e-06, + "loss": 0.0839, "step": 9315 }, { - "epoch": 0.52, - "learning_rate": 9.66627797212505e-06, - "loss": 0.0948, + "epoch": 1.03, + "learning_rate": 9.664593513993782e-06, + "loss": 0.0717, "step": 9320 }, { - "epoch": 0.52, - "learning_rate": 9.660725192959078e-06, - "loss": 0.1171, + "epoch": 1.04, + "learning_rate": 9.659040426477123e-06, + "loss": 0.0909, "step": 9325 }, { - "epoch": 0.52, - "learning_rate": 9.655172413793105e-06, - "loss": 0.0487, + "epoch": 1.04, + "learning_rate": 9.653487338960464e-06, + "loss": 0.0772, "step": 9330 }, { - "epoch": 0.52, - "learning_rate": 9.649619634627132e-06, - "loss": 0.1032, + "epoch": 1.04, + "learning_rate": 9.647934251443803e-06, + "loss": 0.0809, "step": 9335 }, { - "epoch": 0.52, - "learning_rate": 9.644066855461159e-06, - "loss": 0.0804, + "epoch": 1.04, + "learning_rate": 9.642381163927144e-06, + "loss": 0.0698, "step": 9340 }, { - "epoch": 0.52, - "learning_rate": 9.638514076295188e-06, - "loss": 0.3292, + "epoch": 1.04, + "learning_rate": 9.636828076410485e-06, + "loss": 0.0991, "step": 9345 }, { - "epoch": 0.52, - "learning_rate": 9.632961297129214e-06, - "loss": 0.3593, + "epoch": 1.04, + "learning_rate": 9.631274988893826e-06, + "loss": 0.1132, "step": 9350 }, { - "epoch": 0.52, - "learning_rate": 9.627408517963241e-06, - "loss": 0.087, + "epoch": 1.04, + "learning_rate": 9.625721901377166e-06, + "loss": 0.1813, "step": 9355 }, { - "epoch": 0.52, - "learning_rate": 9.621855738797268e-06, - "loss": 0.1178, + "epoch": 1.04, + "learning_rate": 9.620168813860507e-06, + "loss": 0.2971, "step": 9360 }, { - "epoch": 0.52, - "learning_rate": 9.616302959631297e-06, - "loss": 0.1203, + "epoch": 1.04, + "learning_rate": 9.614615726343848e-06, + "loss": 0.0776, "step": 9365 }, { - "epoch": 0.52, - "learning_rate": 9.610750180465324e-06, - "loss": 0.1616, + "epoch": 1.04, + "learning_rate": 9.609062638827189e-06, + "loss": 0.0964, "step": 9370 }, { - "epoch": 0.52, - "learning_rate": 9.605197401299351e-06, - "loss": 0.0739, + "epoch": 1.04, + "learning_rate": 9.60350955131053e-06, + "loss": 0.0816, "step": 9375 }, { - "epoch": 0.52, - "learning_rate": 9.599644622133378e-06, - "loss": 0.0781, + "epoch": 1.04, + "learning_rate": 9.597956463793871e-06, + "loss": 0.1118, "step": 9380 }, { - "epoch": 0.52, - "learning_rate": 9.594091842967407e-06, - "loss": 0.0704, + "epoch": 1.04, + "learning_rate": 9.59240337627721e-06, + "loss": 0.0642, "step": 9385 }, { - "epoch": 0.52, - "learning_rate": 9.588539063801433e-06, - "loss": 0.1076, + "epoch": 1.04, + "learning_rate": 9.586850288760551e-06, + "loss": 0.0904, "step": 9390 }, { - "epoch": 0.52, - "learning_rate": 9.58298628463546e-06, - "loss": 0.2476, + "epoch": 1.04, + "learning_rate": 9.581297201243893e-06, + "loss": 0.1273, "step": 9395 }, { - "epoch": 0.52, - "learning_rate": 9.577433505469487e-06, - "loss": 0.2883, + "epoch": 1.04, + "learning_rate": 9.575744113727234e-06, + "loss": 0.1431, "step": 9400 }, { - "epoch": 0.52, - "learning_rate": 9.571880726303516e-06, - "loss": 0.0796, + "epoch": 1.04, + "learning_rate": 9.570191026210573e-06, + "loss": 0.1998, "step": 9405 }, { - "epoch": 0.52, - "learning_rate": 9.566327947137543e-06, - "loss": 0.116, + "epoch": 1.04, + "learning_rate": 9.564637938693914e-06, + "loss": 0.2178, "step": 9410 }, { - "epoch": 0.52, - "learning_rate": 9.560775167971572e-06, - "loss": 0.1027, + "epoch": 1.05, + "learning_rate": 9.559084851177255e-06, + "loss": 0.0652, "step": 9415 }, { - "epoch": 0.52, - "learning_rate": 9.555222388805599e-06, - "loss": 0.0697, + "epoch": 1.05, + "learning_rate": 9.553531763660596e-06, + "loss": 0.0782, "step": 9420 }, { - "epoch": 0.52, - "learning_rate": 9.549669609639625e-06, - "loss": 0.0751, + "epoch": 1.05, + "learning_rate": 9.547978676143937e-06, + "loss": 0.0775, "step": 9425 }, { - "epoch": 0.52, - "learning_rate": 9.544116830473652e-06, - "loss": 0.0737, + "epoch": 1.05, + "learning_rate": 9.542425588627278e-06, + "loss": 0.0878, "step": 9430 }, { - "epoch": 0.52, - "learning_rate": 9.538564051307681e-06, - "loss": 0.0865, + "epoch": 1.05, + "learning_rate": 9.536872501110618e-06, + "loss": 0.0363, "step": 9435 }, { - "epoch": 0.52, - "learning_rate": 9.533011272141708e-06, - "loss": 0.1445, + "epoch": 1.05, + "learning_rate": 9.531319413593959e-06, + "loss": 0.0745, "step": 9440 }, { - "epoch": 0.52, - "learning_rate": 9.527458492975735e-06, - "loss": 0.1595, + "epoch": 1.05, + "learning_rate": 9.5257663260773e-06, + "loss": 0.0608, "step": 9445 }, { - "epoch": 0.52, - "learning_rate": 9.521905713809762e-06, - "loss": 0.309, + "epoch": 1.05, + "learning_rate": 9.52021323856064e-06, + "loss": 0.1054, "step": 9450 }, { - "epoch": 0.52, - "learning_rate": 9.51635293464379e-06, - "loss": 0.0652, + "epoch": 1.05, + "learning_rate": 9.514660151043982e-06, + "loss": 0.1699, "step": 9455 }, { - "epoch": 0.52, - "learning_rate": 9.510800155477818e-06, - "loss": 0.1278, + "epoch": 1.05, + "learning_rate": 9.509107063527321e-06, + "loss": 0.2034, "step": 9460 }, { - "epoch": 0.53, - "learning_rate": 9.505247376311844e-06, - "loss": 0.1057, + "epoch": 1.05, + "learning_rate": 9.503553976010664e-06, + "loss": 0.0742, "step": 9465 }, { - "epoch": 0.53, - "learning_rate": 9.499694597145871e-06, - "loss": 0.1111, + "epoch": 1.05, + "learning_rate": 9.498000888494003e-06, + "loss": 0.0641, "step": 9470 }, { - "epoch": 0.53, - "learning_rate": 9.4941418179799e-06, - "loss": 0.0922, + "epoch": 1.05, + "learning_rate": 9.492447800977344e-06, + "loss": 0.0714, "step": 9475 }, { - "epoch": 0.53, - "learning_rate": 9.488589038813927e-06, - "loss": 0.0392, + "epoch": 1.05, + "learning_rate": 9.486894713460685e-06, + "loss": 0.1139, "step": 9480 }, { - "epoch": 0.53, - "learning_rate": 9.483036259647954e-06, - "loss": 0.083, + "epoch": 1.05, + "learning_rate": 9.481341625944027e-06, + "loss": 0.0519, "step": 9485 }, { - "epoch": 0.53, - "learning_rate": 9.477483480481981e-06, - "loss": 0.1164, + "epoch": 1.05, + "learning_rate": 9.475788538427366e-06, + "loss": 0.058, "step": 9490 }, { - "epoch": 0.53, - "learning_rate": 9.47193070131601e-06, - "loss": 0.185, + "epoch": 1.05, + "learning_rate": 9.470235450910707e-06, + "loss": 0.0923, "step": 9495 }, { - "epoch": 0.53, - "learning_rate": 9.466377922150038e-06, - "loss": 0.481, + "epoch": 1.05, + "learning_rate": 9.464682363394048e-06, + "loss": 0.115, "step": 9500 }, { - "epoch": 0.53, - "learning_rate": 9.460825142984065e-06, - "loss": 0.115, + "epoch": 1.06, + "learning_rate": 9.459129275877389e-06, + "loss": 0.2176, "step": 9505 }, { - "epoch": 0.53, - "learning_rate": 9.455272363818092e-06, - "loss": 0.1101, + "epoch": 1.06, + "learning_rate": 9.453576188360728e-06, + "loss": 0.2731, "step": 9510 }, { - "epoch": 0.53, - "learning_rate": 9.449719584652119e-06, - "loss": 0.1624, + "epoch": 1.06, + "learning_rate": 9.44802310084407e-06, + "loss": 0.0734, "step": 9515 }, { - "epoch": 0.53, - "learning_rate": 9.444166805486148e-06, - "loss": 0.1529, + "epoch": 1.06, + "learning_rate": 9.44247001332741e-06, + "loss": 0.0653, "step": 9520 }, { - "epoch": 0.53, - "learning_rate": 9.438614026320175e-06, - "loss": 0.0945, + "epoch": 1.06, + "learning_rate": 9.436916925810752e-06, + "loss": 0.0732, "step": 9525 }, { - "epoch": 0.53, - "learning_rate": 9.433061247154202e-06, - "loss": 0.1185, + "epoch": 1.06, + "learning_rate": 9.431363838294093e-06, + "loss": 0.0794, "step": 9530 }, { - "epoch": 0.53, - "learning_rate": 9.427508467988229e-06, - "loss": 0.1339, + "epoch": 1.06, + "learning_rate": 9.425810750777434e-06, + "loss": 0.068, "step": 9535 }, { - "epoch": 0.53, - "learning_rate": 9.421955688822255e-06, - "loss": 0.1046, + "epoch": 1.06, + "learning_rate": 9.420257663260773e-06, + "loss": 0.0973, "step": 9540 }, { - "epoch": 0.53, - "learning_rate": 9.416402909656284e-06, - "loss": 0.1605, + "epoch": 1.06, + "learning_rate": 9.414704575744114e-06, + "loss": 0.1599, "step": 9545 }, { - "epoch": 0.53, - "learning_rate": 9.410850130490311e-06, - "loss": 0.2686, + "epoch": 1.06, + "learning_rate": 9.409151488227455e-06, + "loss": 0.1473, "step": 9550 }, { - "epoch": 0.53, - "learning_rate": 9.405297351324338e-06, - "loss": 0.0831, + "epoch": 1.06, + "learning_rate": 9.403598400710796e-06, + "loss": 0.1827, "step": 9555 }, { - "epoch": 0.53, - "learning_rate": 9.399744572158365e-06, - "loss": 0.1146, + "epoch": 1.06, + "learning_rate": 9.398045313194136e-06, + "loss": 0.2009, "step": 9560 }, { - "epoch": 0.53, - "learning_rate": 9.394191792992394e-06, - "loss": 0.1213, + "epoch": 1.06, + "learning_rate": 9.392492225677477e-06, + "loss": 0.0616, "step": 9565 }, { - "epoch": 0.53, - "learning_rate": 9.38863901382642e-06, - "loss": 0.1401, + "epoch": 1.06, + "learning_rate": 9.386939138160818e-06, + "loss": 0.0808, "step": 9570 }, { - "epoch": 0.53, - "learning_rate": 9.383086234660448e-06, - "loss": 0.11, + "epoch": 1.06, + "learning_rate": 9.381386050644159e-06, + "loss": 0.0805, "step": 9575 }, { - "epoch": 0.53, - "learning_rate": 9.377533455494474e-06, - "loss": 0.0404, + "epoch": 1.06, + "learning_rate": 9.3758329631275e-06, + "loss": 0.0839, "step": 9580 }, { - "epoch": 0.53, - "learning_rate": 9.371980676328503e-06, - "loss": 0.0817, + "epoch": 1.06, + "learning_rate": 9.370279875610841e-06, + "loss": 0.0395, "step": 9585 }, { - "epoch": 0.53, - "learning_rate": 9.366427897162532e-06, - "loss": 0.1301, + "epoch": 1.06, + "learning_rate": 9.364726788094182e-06, + "loss": 0.0223, "step": 9590 }, { - "epoch": 0.53, - "learning_rate": 9.360875117996559e-06, - "loss": 0.1596, + "epoch": 1.07, + "learning_rate": 9.359173700577521e-06, + "loss": 0.0583, "step": 9595 }, { - "epoch": 0.53, - "learning_rate": 9.355322338830586e-06, - "loss": 0.4307, + "epoch": 1.07, + "learning_rate": 9.353620613060863e-06, + "loss": 0.0823, "step": 9600 }, { - "epoch": 0.53, - "learning_rate": 9.349769559664613e-06, - "loss": 0.1022, + "epoch": 1.07, + "learning_rate": 9.348067525544204e-06, + "loss": 0.138, "step": 9605 }, { - "epoch": 0.53, - "learning_rate": 9.344216780498641e-06, - "loss": 0.0639, + "epoch": 1.07, + "learning_rate": 9.342514438027545e-06, + "loss": 0.3078, "step": 9610 }, { - "epoch": 0.53, - "learning_rate": 9.338664001332668e-06, - "loss": 0.1554, + "epoch": 1.07, + "learning_rate": 9.336961350510884e-06, + "loss": 0.0712, "step": 9615 }, { - "epoch": 0.53, - "learning_rate": 9.333111222166695e-06, - "loss": 0.1479, + "epoch": 1.07, + "learning_rate": 9.331408262994227e-06, + "loss": 0.0754, "step": 9620 }, { - "epoch": 0.53, - "learning_rate": 9.327558443000722e-06, - "loss": 0.0546, + "epoch": 1.07, + "learning_rate": 9.325855175477566e-06, + "loss": 0.0889, "step": 9625 }, { - "epoch": 0.53, - "learning_rate": 9.32200566383475e-06, - "loss": 0.0552, + "epoch": 1.07, + "learning_rate": 9.320302087960907e-06, + "loss": 0.1146, "step": 9630 }, { - "epoch": 0.53, - "learning_rate": 9.316452884668778e-06, - "loss": 0.1299, + "epoch": 1.07, + "learning_rate": 9.314749000444248e-06, + "loss": 0.0488, "step": 9635 }, { - "epoch": 0.53, - "learning_rate": 9.310900105502805e-06, - "loss": 0.178, + "epoch": 1.07, + "learning_rate": 9.30919591292759e-06, + "loss": 0.0445, "step": 9640 }, { - "epoch": 0.54, - "learning_rate": 9.305347326336832e-06, - "loss": 0.2167, + "epoch": 1.07, + "learning_rate": 9.303642825410929e-06, + "loss": 0.0777, "step": 9645 }, { - "epoch": 0.54, - "learning_rate": 9.29979454717086e-06, - "loss": 0.3389, + "epoch": 1.07, + "learning_rate": 9.29808973789427e-06, + "loss": 0.1293, "step": 9650 }, { - "epoch": 0.54, - "learning_rate": 9.294241768004887e-06, - "loss": 0.1231, + "epoch": 1.07, + "learning_rate": 9.29253665037761e-06, + "loss": 0.1418, "step": 9655 }, { - "epoch": 0.54, - "learning_rate": 9.288688988838914e-06, - "loss": 0.0984, + "epoch": 1.07, + "learning_rate": 9.286983562860952e-06, + "loss": 0.2381, "step": 9660 }, { - "epoch": 0.54, - "learning_rate": 9.283136209672941e-06, - "loss": 0.0783, + "epoch": 1.07, + "learning_rate": 9.281430475344291e-06, + "loss": 0.0638, "step": 9665 }, { - "epoch": 0.54, - "learning_rate": 9.27758343050697e-06, - "loss": 0.1135, + "epoch": 1.07, + "learning_rate": 9.275877387827632e-06, + "loss": 0.0684, "step": 9670 }, { - "epoch": 0.54, - "learning_rate": 9.272030651340997e-06, - "loss": 0.0975, + "epoch": 1.07, + "learning_rate": 9.270324300310973e-06, + "loss": 0.0814, "step": 9675 }, { - "epoch": 0.54, - "learning_rate": 9.266477872175025e-06, - "loss": 0.0512, + "epoch": 1.07, + "learning_rate": 9.264771212794314e-06, + "loss": 0.1062, "step": 9680 }, { - "epoch": 0.54, - "learning_rate": 9.260925093009052e-06, - "loss": 0.1254, + "epoch": 1.08, + "learning_rate": 9.259218125277655e-06, + "loss": 0.0433, "step": 9685 }, { - "epoch": 0.54, - "learning_rate": 9.25537231384308e-06, - "loss": 0.1406, + "epoch": 1.08, + "learning_rate": 9.253665037760997e-06, + "loss": 0.049, "step": 9690 }, { - "epoch": 0.54, - "learning_rate": 9.249819534677106e-06, - "loss": 0.124, + "epoch": 1.08, + "learning_rate": 9.248111950244336e-06, + "loss": 0.0646, "step": 9695 }, { - "epoch": 0.54, - "learning_rate": 9.244266755511135e-06, - "loss": 0.2948, + "epoch": 1.08, + "learning_rate": 9.242558862727677e-06, + "loss": 0.1562, "step": 9700 }, { - "epoch": 0.54, - "learning_rate": 9.238713976345162e-06, - "loss": 0.1102, + "epoch": 1.08, + "learning_rate": 9.237005775211018e-06, + "loss": 0.1716, "step": 9705 }, { - "epoch": 0.54, - "learning_rate": 9.233161197179189e-06, - "loss": 0.0842, + "epoch": 1.08, + "learning_rate": 9.231452687694359e-06, + "loss": 0.2675, "step": 9710 }, { - "epoch": 0.54, - "learning_rate": 9.227608418013216e-06, - "loss": 0.1301, + "epoch": 1.08, + "learning_rate": 9.225899600177698e-06, + "loss": 0.076, "step": 9715 }, { - "epoch": 0.54, - "learning_rate": 9.222055638847244e-06, - "loss": 0.1238, + "epoch": 1.08, + "learning_rate": 9.22034651266104e-06, + "loss": 0.0608, "step": 9720 }, { - "epoch": 0.54, - "learning_rate": 9.216502859681271e-06, - "loss": 0.1007, + "epoch": 1.08, + "learning_rate": 9.214793425144382e-06, + "loss": 0.0845, "step": 9725 }, { - "epoch": 0.54, - "learning_rate": 9.210950080515298e-06, - "loss": 0.0747, + "epoch": 1.08, + "learning_rate": 9.209240337627722e-06, + "loss": 0.0688, "step": 9730 }, { - "epoch": 0.54, - "learning_rate": 9.205397301349325e-06, - "loss": 0.1378, + "epoch": 1.08, + "learning_rate": 9.203687250111063e-06, + "loss": 0.0523, "step": 9735 }, { - "epoch": 0.54, - "learning_rate": 9.199844522183354e-06, - "loss": 0.0679, + "epoch": 1.08, + "learning_rate": 9.198134162594404e-06, + "loss": 0.0277, "step": 9740 }, { - "epoch": 0.54, - "learning_rate": 9.19429174301738e-06, - "loss": 0.1749, + "epoch": 1.08, + "learning_rate": 9.192581075077745e-06, + "loss": 0.0661, "step": 9745 }, { - "epoch": 0.54, - "learning_rate": 9.188738963851408e-06, - "loss": 0.3752, + "epoch": 1.08, + "learning_rate": 9.187027987561084e-06, + "loss": 0.1164, "step": 9750 }, { - "epoch": 0.54, - "learning_rate": 9.183186184685435e-06, - "loss": 0.1076, + "epoch": 1.08, + "learning_rate": 9.181474900044425e-06, + "loss": 0.111, "step": 9755 }, { - "epoch": 0.54, - "learning_rate": 9.177633405519463e-06, - "loss": 0.1079, + "epoch": 1.08, + "learning_rate": 9.175921812527766e-06, + "loss": 0.2074, "step": 9760 }, { - "epoch": 0.54, - "learning_rate": 9.172080626353492e-06, - "loss": 0.0925, + "epoch": 1.08, + "learning_rate": 9.170368725011107e-06, + "loss": 0.0627, "step": 9765 }, { - "epoch": 0.54, - "learning_rate": 9.166527847187519e-06, - "loss": 0.1112, + "epoch": 1.08, + "learning_rate": 9.164815637494447e-06, + "loss": 0.132, "step": 9770 }, { - "epoch": 0.54, - "learning_rate": 9.160975068021546e-06, - "loss": 0.1049, + "epoch": 1.09, + "learning_rate": 9.15926254997779e-06, + "loss": 0.0857, "step": 9775 }, { - "epoch": 0.54, - "learning_rate": 9.155422288855573e-06, - "loss": 0.0674, + "epoch": 1.09, + "learning_rate": 9.153709462461129e-06, + "loss": 0.0661, "step": 9780 }, { - "epoch": 0.54, - "learning_rate": 9.149869509689601e-06, - "loss": 0.0825, + "epoch": 1.09, + "learning_rate": 9.14815637494447e-06, + "loss": 0.1088, "step": 9785 }, { - "epoch": 0.54, - "learning_rate": 9.144316730523628e-06, - "loss": 0.103, + "epoch": 1.09, + "learning_rate": 9.142603287427811e-06, + "loss": 0.0513, "step": 9790 }, { - "epoch": 0.54, - "learning_rate": 9.138763951357655e-06, - "loss": 0.2101, + "epoch": 1.09, + "learning_rate": 9.137050199911152e-06, + "loss": 0.0719, "step": 9795 }, { - "epoch": 0.54, - "learning_rate": 9.133211172191682e-06, - "loss": 0.4221, + "epoch": 1.09, + "learning_rate": 9.131497112394491e-06, + "loss": 0.07, "step": 9800 }, { - "epoch": 0.54, - "learning_rate": 9.127658393025711e-06, - "loss": 0.1155, + "epoch": 1.09, + "learning_rate": 9.125944024877833e-06, + "loss": 0.1544, "step": 9805 }, { - "epoch": 0.54, - "learning_rate": 9.122105613859738e-06, - "loss": 0.1017, + "epoch": 1.09, + "learning_rate": 9.120390937361174e-06, + "loss": 0.2289, "step": 9810 }, { - "epoch": 0.54, - "learning_rate": 9.116552834693765e-06, - "loss": 0.1237, + "epoch": 1.09, + "learning_rate": 9.114837849844515e-06, + "loss": 0.0585, "step": 9815 }, { - "epoch": 0.54, - "learning_rate": 9.111000055527792e-06, - "loss": 0.116, + "epoch": 1.09, + "learning_rate": 9.109284762327854e-06, + "loss": 0.0721, "step": 9820 }, { - "epoch": 0.55, - "learning_rate": 9.10544727636182e-06, - "loss": 0.0887, + "epoch": 1.09, + "learning_rate": 9.103731674811195e-06, + "loss": 0.0831, "step": 9825 }, { - "epoch": 0.55, - "learning_rate": 9.099894497195847e-06, - "loss": 0.0525, + "epoch": 1.09, + "learning_rate": 9.098178587294536e-06, + "loss": 0.0853, "step": 9830 }, { - "epoch": 0.55, - "learning_rate": 9.094341718029874e-06, - "loss": 0.1004, + "epoch": 1.09, + "learning_rate": 9.092625499777877e-06, + "loss": 0.0464, "step": 9835 }, { - "epoch": 0.55, - "learning_rate": 9.088788938863901e-06, - "loss": 0.0867, + "epoch": 1.09, + "learning_rate": 9.087072412261218e-06, + "loss": 0.0464, "step": 9840 }, { - "epoch": 0.55, - "learning_rate": 9.08323615969793e-06, - "loss": 0.1833, + "epoch": 1.09, + "learning_rate": 9.08151932474456e-06, + "loss": 0.0639, "step": 9845 }, { - "epoch": 0.55, - "learning_rate": 9.077683380531957e-06, - "loss": 0.2944, + "epoch": 1.09, + "learning_rate": 9.075966237227899e-06, + "loss": 0.0548, "step": 9850 }, { - "epoch": 0.55, - "learning_rate": 9.072130601365985e-06, - "loss": 0.1141, + "epoch": 1.09, + "learning_rate": 9.07041314971124e-06, + "loss": 0.196, "step": 9855 }, { - "epoch": 0.55, - "learning_rate": 9.066577822200012e-06, - "loss": 0.1039, + "epoch": 1.09, + "learning_rate": 9.06486006219458e-06, + "loss": 0.2379, "step": 9860 }, { - "epoch": 0.55, - "learning_rate": 9.06102504303404e-06, - "loss": 0.0802, + "epoch": 1.1, + "learning_rate": 9.059306974677922e-06, + "loss": 0.0846, "step": 9865 }, { - "epoch": 0.55, - "learning_rate": 9.055472263868066e-06, - "loss": 0.1491, + "epoch": 1.1, + "learning_rate": 9.053753887161263e-06, + "loss": 0.0647, "step": 9870 }, { - "epoch": 0.55, - "learning_rate": 9.049919484702095e-06, - "loss": 0.1094, + "epoch": 1.1, + "learning_rate": 9.048200799644602e-06, + "loss": 0.098, "step": 9875 }, { - "epoch": 0.55, - "learning_rate": 9.044366705536122e-06, - "loss": 0.0701, + "epoch": 1.1, + "learning_rate": 9.042647712127945e-06, + "loss": 0.0939, "step": 9880 }, { - "epoch": 0.55, - "learning_rate": 9.038813926370149e-06, - "loss": 0.2822, + "epoch": 1.1, + "learning_rate": 9.037094624611284e-06, + "loss": 0.0553, "step": 9885 }, { - "epoch": 0.55, - "learning_rate": 9.033261147204176e-06, - "loss": 0.1946, + "epoch": 1.1, + "learning_rate": 9.031541537094626e-06, + "loss": 0.0505, "step": 9890 }, { - "epoch": 0.55, - "learning_rate": 9.027708368038204e-06, - "loss": 0.1355, + "epoch": 1.1, + "learning_rate": 9.025988449577967e-06, + "loss": 0.0596, "step": 9895 }, { - "epoch": 0.55, - "learning_rate": 9.022155588872231e-06, - "loss": 0.5325, + "epoch": 1.1, + "learning_rate": 9.020435362061308e-06, + "loss": 0.097, "step": 9900 }, { - "epoch": 0.55, - "learning_rate": 9.016602809706258e-06, - "loss": 0.0848, + "epoch": 1.1, + "learning_rate": 9.014882274544647e-06, + "loss": 0.1881, "step": 9905 }, { - "epoch": 0.55, - "learning_rate": 9.011050030540285e-06, - "loss": 0.0871, + "epoch": 1.1, + "learning_rate": 9.009329187027988e-06, + "loss": 0.2562, "step": 9910 }, { - "epoch": 0.55, - "learning_rate": 9.005497251374314e-06, - "loss": 0.1102, + "epoch": 1.1, + "learning_rate": 9.003776099511329e-06, + "loss": 0.0614, "step": 9915 }, { - "epoch": 0.55, - "learning_rate": 8.999944472208341e-06, - "loss": 0.1492, + "epoch": 1.1, + "learning_rate": 8.99822301199467e-06, + "loss": 0.0803, "step": 9920 }, { - "epoch": 0.55, - "learning_rate": 8.994391693042368e-06, - "loss": 0.0725, + "epoch": 1.1, + "learning_rate": 8.99266992447801e-06, + "loss": 0.0694, "step": 9925 }, { - "epoch": 0.55, - "learning_rate": 8.988838913876395e-06, - "loss": 0.0408, + "epoch": 1.1, + "learning_rate": 8.987116836961352e-06, + "loss": 0.0807, "step": 9930 }, { - "epoch": 0.55, - "learning_rate": 8.983286134710423e-06, - "loss": 0.0904, + "epoch": 1.1, + "learning_rate": 8.981563749444692e-06, + "loss": 0.0658, "step": 9935 }, { - "epoch": 0.55, - "learning_rate": 8.977733355544452e-06, - "loss": 0.199, + "epoch": 1.1, + "learning_rate": 8.976010661928033e-06, + "loss": 0.0545, "step": 9940 }, { - "epoch": 0.55, - "learning_rate": 8.972180576378479e-06, - "loss": 0.167, + "epoch": 1.1, + "learning_rate": 8.970457574411374e-06, + "loss": 0.0712, "step": 9945 }, { - "epoch": 0.55, - "learning_rate": 8.966627797212506e-06, - "loss": 0.2511, + "epoch": 1.1, + "learning_rate": 8.964904486894715e-06, + "loss": 0.0791, "step": 9950 }, { - "epoch": 0.55, - "learning_rate": 8.961075018046533e-06, - "loss": 0.1001, + "epoch": 1.1, + "learning_rate": 8.959351399378054e-06, + "loss": 0.1138, "step": 9955 }, { - "epoch": 0.55, - "learning_rate": 8.955522238880562e-06, - "loss": 0.1388, + "epoch": 1.11, + "learning_rate": 8.953798311861395e-06, + "loss": 0.2095, "step": 9960 }, { - "epoch": 0.55, - "learning_rate": 8.949969459714588e-06, - "loss": 0.1363, + "epoch": 1.11, + "learning_rate": 8.948245224344736e-06, + "loss": 0.0778, "step": 9965 }, { - "epoch": 0.55, - "learning_rate": 8.944416680548615e-06, - "loss": 0.1009, + "epoch": 1.11, + "learning_rate": 8.942692136828077e-06, + "loss": 0.0691, "step": 9970 }, { - "epoch": 0.55, - "learning_rate": 8.938863901382642e-06, - "loss": 0.0778, + "epoch": 1.11, + "learning_rate": 8.937139049311417e-06, + "loss": 0.0888, "step": 9975 }, { - "epoch": 0.55, - "learning_rate": 8.933311122216671e-06, - "loss": 0.0765, + "epoch": 1.11, + "learning_rate": 8.931585961794758e-06, + "loss": 0.1236, "step": 9980 }, { - "epoch": 0.55, - "learning_rate": 8.927758343050698e-06, - "loss": 0.0988, + "epoch": 1.11, + "learning_rate": 8.926032874278099e-06, + "loss": 0.0293, "step": 9985 }, { - "epoch": 0.55, - "learning_rate": 8.922205563884725e-06, - "loss": 0.1303, + "epoch": 1.11, + "learning_rate": 8.92047978676144e-06, + "loss": 0.0301, "step": 9990 }, { - "epoch": 0.55, - "learning_rate": 8.916652784718752e-06, - "loss": 0.1803, + "epoch": 1.11, + "learning_rate": 8.914926699244781e-06, + "loss": 0.057, "step": 9995 }, { - "epoch": 0.55, - "learning_rate": 8.911100005552779e-06, - "loss": 0.3222, + "epoch": 1.11, + "learning_rate": 8.909373611728122e-06, + "loss": 0.083, "step": 10000 }, { - "epoch": 0.55, - "eval_loss": 0.050108496099710464, - "eval_runtime": 1614.6557, - "eval_samples_per_second": 9.92, - "eval_steps_per_second": 2.48, - "eval_wer": 0.13211323476379372, + "epoch": 1.11, + "eval_loss": 0.033624935895204544, + "eval_runtime": 1824.5042, + "eval_samples_per_second": 8.779, + "eval_steps_per_second": 2.195, + "eval_wer": 0.08218535881716553, "step": 10000 }, { - "epoch": 0.56, - "learning_rate": 8.905547226386807e-06, - "loss": 0.0772, + "epoch": 1.11, + "learning_rate": 8.903820524211463e-06, + "loss": 0.1404, "step": 10005 }, { - "epoch": 0.56, - "learning_rate": 8.899994447220834e-06, - "loss": 0.1281, + "epoch": 1.11, + "learning_rate": 8.898267436694803e-06, + "loss": 0.2263, "step": 10010 }, { - "epoch": 0.56, - "learning_rate": 8.894441668054861e-06, - "loss": 0.0962, + "epoch": 1.11, + "learning_rate": 8.892714349178144e-06, + "loss": 0.0728, "step": 10015 }, { - "epoch": 0.56, - "learning_rate": 8.888888888888888e-06, - "loss": 0.1088, + "epoch": 1.11, + "learning_rate": 8.887161261661485e-06, + "loss": 0.0705, "step": 10020 }, { - "epoch": 0.56, - "learning_rate": 8.883336109722917e-06, - "loss": 0.1011, + "epoch": 1.11, + "learning_rate": 8.881608174144826e-06, + "loss": 0.0972, "step": 10025 }, { - "epoch": 0.56, - "learning_rate": 8.877783330556946e-06, - "loss": 0.148, + "epoch": 1.11, + "learning_rate": 8.876055086628165e-06, + "loss": 0.0948, "step": 10030 }, { - "epoch": 0.56, - "learning_rate": 8.872230551390973e-06, - "loss": 0.103, + "epoch": 1.11, + "learning_rate": 8.870501999111508e-06, + "loss": 0.0245, "step": 10035 }, { - "epoch": 0.56, - "learning_rate": 8.866677772225e-06, - "loss": 0.2297, + "epoch": 1.11, + "learning_rate": 8.864948911594847e-06, + "loss": 0.0558, "step": 10040 }, { - "epoch": 0.56, - "learning_rate": 8.861124993059026e-06, - "loss": 0.2146, + "epoch": 1.11, + "learning_rate": 8.859395824078188e-06, + "loss": 0.0816, "step": 10045 }, { - "epoch": 0.56, - "learning_rate": 8.855572213893055e-06, - "loss": 0.2865, + "epoch": 1.12, + "learning_rate": 8.85384273656153e-06, + "loss": 0.1294, "step": 10050 }, { - "epoch": 0.56, - "learning_rate": 8.850019434727082e-06, - "loss": 0.1353, + "epoch": 1.12, + "learning_rate": 8.84828964904487e-06, + "loss": 0.2248, "step": 10055 }, { - "epoch": 0.56, - "learning_rate": 8.844466655561109e-06, - "loss": 0.1243, + "epoch": 1.12, + "learning_rate": 8.84273656152821e-06, + "loss": 0.2146, "step": 10060 }, { - "epoch": 0.56, - "learning_rate": 8.838913876395136e-06, - "loss": 0.1072, + "epoch": 1.12, + "learning_rate": 8.83718347401155e-06, + "loss": 0.0828, "step": 10065 }, { - "epoch": 0.56, - "learning_rate": 8.833361097229165e-06, - "loss": 0.145, + "epoch": 1.12, + "learning_rate": 8.831630386494892e-06, + "loss": 0.0751, "step": 10070 }, { - "epoch": 0.56, - "learning_rate": 8.827808318063192e-06, - "loss": 0.094, + "epoch": 1.12, + "learning_rate": 8.826077298978233e-06, + "loss": 0.0982, "step": 10075 }, { - "epoch": 0.56, - "learning_rate": 8.822255538897218e-06, - "loss": 0.0823, + "epoch": 1.12, + "learning_rate": 8.820524211461572e-06, + "loss": 0.1165, "step": 10080 }, { - "epoch": 0.56, - "learning_rate": 8.816702759731245e-06, - "loss": 0.1158, + "epoch": 1.12, + "learning_rate": 8.814971123944915e-06, + "loss": 0.0855, "step": 10085 }, { - "epoch": 0.56, - "learning_rate": 8.811149980565274e-06, - "loss": 0.1, + "epoch": 1.12, + "learning_rate": 8.809418036428254e-06, + "loss": 0.0588, "step": 10090 }, { - "epoch": 0.56, - "learning_rate": 8.805597201399301e-06, - "loss": 0.2373, + "epoch": 1.12, + "learning_rate": 8.803864948911596e-06, + "loss": 0.0708, "step": 10095 }, { - "epoch": 0.56, - "learning_rate": 8.800044422233328e-06, - "loss": 0.5565, + "epoch": 1.12, + "learning_rate": 8.798311861394937e-06, + "loss": 0.1188, "step": 10100 }, { - "epoch": 0.56, - "learning_rate": 8.794491643067355e-06, - "loss": 0.0971, + "epoch": 1.12, + "learning_rate": 8.792758773878278e-06, + "loss": 0.1458, "step": 10105 }, { - "epoch": 0.56, - "learning_rate": 8.788938863901384e-06, - "loss": 0.0965, + "epoch": 1.12, + "learning_rate": 8.787205686361617e-06, + "loss": 0.3549, "step": 10110 }, { - "epoch": 0.56, - "learning_rate": 8.78338608473541e-06, - "loss": 0.1217, + "epoch": 1.12, + "learning_rate": 8.781652598844958e-06, + "loss": 0.051, "step": 10115 }, { - "epoch": 0.56, - "learning_rate": 8.777833305569439e-06, - "loss": 0.1035, + "epoch": 1.12, + "learning_rate": 8.776099511328299e-06, + "loss": 0.0744, "step": 10120 }, { - "epoch": 0.56, - "learning_rate": 8.772280526403466e-06, - "loss": 0.1477, + "epoch": 1.12, + "learning_rate": 8.77054642381164e-06, + "loss": 0.0744, "step": 10125 }, { - "epoch": 0.56, - "learning_rate": 8.766727747237493e-06, - "loss": 0.0202, + "epoch": 1.12, + "learning_rate": 8.764993336294981e-06, + "loss": 0.0964, "step": 10130 }, { - "epoch": 0.56, - "learning_rate": 8.76117496807152e-06, - "loss": 0.0798, + "epoch": 1.12, + "learning_rate": 8.75944024877832e-06, + "loss": 0.0551, "step": 10135 }, { - "epoch": 0.56, - "learning_rate": 8.755622188905549e-06, - "loss": 0.1072, + "epoch": 1.13, + "learning_rate": 8.753887161261663e-06, + "loss": 0.041, "step": 10140 }, { - "epoch": 0.56, - "learning_rate": 8.750069409739576e-06, - "loss": 0.253, + "epoch": 1.13, + "learning_rate": 8.748334073745003e-06, + "loss": 0.0542, "step": 10145 }, { - "epoch": 0.56, - "learning_rate": 8.744516630573603e-06, - "loss": 0.3637, + "epoch": 1.13, + "learning_rate": 8.742780986228344e-06, + "loss": 0.1429, "step": 10150 }, { - "epoch": 0.56, - "learning_rate": 8.73896385140763e-06, - "loss": 0.0748, + "epoch": 1.13, + "learning_rate": 8.737227898711685e-06, + "loss": 0.1656, "step": 10155 }, { - "epoch": 0.56, - "learning_rate": 8.733411072241658e-06, - "loss": 0.1135, + "epoch": 1.13, + "learning_rate": 8.731674811195026e-06, + "loss": 0.242, "step": 10160 }, { - "epoch": 0.56, - "learning_rate": 8.727858293075685e-06, - "loss": 0.1228, + "epoch": 1.13, + "learning_rate": 8.726121723678365e-06, + "loss": 0.0652, "step": 10165 }, { - "epoch": 0.56, - "learning_rate": 8.722305513909712e-06, - "loss": 0.1192, + "epoch": 1.13, + "learning_rate": 8.720568636161706e-06, + "loss": 0.0791, "step": 10170 }, { - "epoch": 0.56, - "learning_rate": 8.716752734743739e-06, - "loss": 0.0932, + "epoch": 1.13, + "learning_rate": 8.715015548645047e-06, + "loss": 0.0712, "step": 10175 }, { - "epoch": 0.56, - "learning_rate": 8.711199955577768e-06, - "loss": 0.0499, + "epoch": 1.13, + "learning_rate": 8.709462461128388e-06, + "loss": 0.1123, "step": 10180 }, { - "epoch": 0.57, - "learning_rate": 8.705647176411795e-06, - "loss": 0.0688, + "epoch": 1.13, + "learning_rate": 8.703909373611728e-06, + "loss": 0.0367, "step": 10185 }, { - "epoch": 0.57, - "learning_rate": 8.700094397245821e-06, - "loss": 0.1914, + "epoch": 1.13, + "learning_rate": 8.69835628609507e-06, + "loss": 0.0754, "step": 10190 }, { - "epoch": 0.57, - "learning_rate": 8.694541618079848e-06, - "loss": 0.1682, + "epoch": 1.13, + "learning_rate": 8.69280319857841e-06, + "loss": 0.0435, "step": 10195 }, { - "epoch": 0.57, - "learning_rate": 8.688988838913877e-06, - "loss": 0.3857, + "epoch": 1.13, + "learning_rate": 8.687250111061751e-06, + "loss": 0.1014, "step": 10200 }, { - "epoch": 0.57, - "learning_rate": 8.683436059747906e-06, - "loss": 0.0819, + "epoch": 1.13, + "learning_rate": 8.681697023545092e-06, + "loss": 0.1813, "step": 10205 }, { - "epoch": 0.57, - "learning_rate": 8.677883280581933e-06, - "loss": 0.1111, + "epoch": 1.13, + "learning_rate": 8.676143936028433e-06, + "loss": 0.2386, "step": 10210 }, { - "epoch": 0.57, - "learning_rate": 8.67233050141596e-06, - "loss": 0.1162, + "epoch": 1.13, + "learning_rate": 8.670590848511773e-06, + "loss": 0.0901, "step": 10215 }, { - "epoch": 0.57, - "learning_rate": 8.666777722249987e-06, - "loss": 0.1903, + "epoch": 1.13, + "learning_rate": 8.665037760995114e-06, + "loss": 0.0924, "step": 10220 }, { - "epoch": 0.57, - "learning_rate": 8.661224943084015e-06, - "loss": 0.104, + "epoch": 1.13, + "learning_rate": 8.659484673478455e-06, + "loss": 0.0571, "step": 10225 }, { - "epoch": 0.57, - "learning_rate": 8.655672163918042e-06, - "loss": 0.1689, + "epoch": 1.14, + "learning_rate": 8.653931585961796e-06, + "loss": 0.0835, "step": 10230 }, { - "epoch": 0.57, - "learning_rate": 8.650119384752069e-06, - "loss": 0.1233, + "epoch": 1.14, + "learning_rate": 8.648378498445135e-06, + "loss": 0.0545, "step": 10235 }, { - "epoch": 0.57, - "learning_rate": 8.644566605586096e-06, - "loss": 0.1582, + "epoch": 1.14, + "learning_rate": 8.642825410928478e-06, + "loss": 0.022, "step": 10240 }, { - "epoch": 0.57, - "learning_rate": 8.639013826420125e-06, - "loss": 0.086, + "epoch": 1.14, + "learning_rate": 8.637272323411817e-06, + "loss": 0.0543, "step": 10245 }, { - "epoch": 0.57, - "learning_rate": 8.633461047254152e-06, - "loss": 0.3574, + "epoch": 1.14, + "learning_rate": 8.631719235895158e-06, + "loss": 0.0893, "step": 10250 }, { - "epoch": 0.57, - "learning_rate": 8.627908268088179e-06, - "loss": 0.0989, + "epoch": 1.14, + "learning_rate": 8.6261661483785e-06, + "loss": 0.1344, "step": 10255 }, { - "epoch": 0.57, - "learning_rate": 8.622355488922206e-06, - "loss": 0.1023, + "epoch": 1.14, + "learning_rate": 8.62061306086184e-06, + "loss": 0.2911, "step": 10260 }, { - "epoch": 0.57, - "learning_rate": 8.616802709756234e-06, - "loss": 0.1227, + "epoch": 1.14, + "learning_rate": 8.61505997334518e-06, + "loss": 0.0561, "step": 10265 }, { - "epoch": 0.57, - "learning_rate": 8.611249930590261e-06, - "loss": 0.1072, + "epoch": 1.14, + "learning_rate": 8.60950688582852e-06, + "loss": 0.0803, "step": 10270 }, { - "epoch": 0.57, - "learning_rate": 8.605697151424288e-06, - "loss": 0.1068, + "epoch": 1.14, + "learning_rate": 8.603953798311862e-06, + "loss": 0.0646, "step": 10275 }, { - "epoch": 0.57, - "learning_rate": 8.600144372258315e-06, - "loss": 0.0497, + "epoch": 1.14, + "learning_rate": 8.598400710795203e-06, + "loss": 0.0943, "step": 10280 }, { - "epoch": 0.57, - "learning_rate": 8.594591593092344e-06, - "loss": 0.0727, + "epoch": 1.14, + "learning_rate": 8.592847623278544e-06, + "loss": 0.055, "step": 10285 }, { - "epoch": 0.57, - "learning_rate": 8.58903881392637e-06, - "loss": 0.1273, + "epoch": 1.14, + "learning_rate": 8.587294535761883e-06, + "loss": 0.0501, "step": 10290 }, { - "epoch": 0.57, - "learning_rate": 8.5834860347604e-06, - "loss": 0.1276, + "epoch": 1.14, + "learning_rate": 8.581741448245226e-06, + "loss": 0.0944, "step": 10295 }, { - "epoch": 0.57, - "learning_rate": 8.577933255594426e-06, - "loss": 0.2308, + "epoch": 1.14, + "learning_rate": 8.576188360728566e-06, + "loss": 0.0865, "step": 10300 }, { - "epoch": 0.57, - "learning_rate": 8.572380476428453e-06, - "loss": 0.1176, + "epoch": 1.14, + "learning_rate": 8.570635273211907e-06, + "loss": 0.1229, "step": 10305 }, { - "epoch": 0.57, - "learning_rate": 8.56682769726248e-06, - "loss": 0.0895, + "epoch": 1.14, + "learning_rate": 8.565082185695248e-06, + "loss": 0.2852, "step": 10310 }, { - "epoch": 0.57, - "learning_rate": 8.561274918096509e-06, - "loss": 0.1006, + "epoch": 1.14, + "learning_rate": 8.559529098178589e-06, + "loss": 0.0619, "step": 10315 }, { - "epoch": 0.57, - "learning_rate": 8.555722138930536e-06, - "loss": 0.0985, + "epoch": 1.15, + "learning_rate": 8.553976010661928e-06, + "loss": 0.074, "step": 10320 }, { - "epoch": 0.57, - "learning_rate": 8.550169359764563e-06, - "loss": 0.1262, + "epoch": 1.15, + "learning_rate": 8.548422923145269e-06, + "loss": 0.0861, "step": 10325 }, { - "epoch": 0.57, - "learning_rate": 8.54461658059859e-06, - "loss": 0.0529, + "epoch": 1.15, + "learning_rate": 8.54286983562861e-06, + "loss": 0.0899, "step": 10330 }, { - "epoch": 0.57, - "learning_rate": 8.539063801432618e-06, - "loss": 0.0498, + "epoch": 1.15, + "learning_rate": 8.537316748111951e-06, + "loss": 0.0766, "step": 10335 }, { - "epoch": 0.57, - "learning_rate": 8.533511022266645e-06, - "loss": 0.1156, + "epoch": 1.15, + "learning_rate": 8.53176366059529e-06, + "loss": 0.0935, "step": 10340 }, { - "epoch": 0.57, - "learning_rate": 8.527958243100672e-06, - "loss": 0.1946, + "epoch": 1.15, + "learning_rate": 8.526210573078633e-06, + "loss": 0.0629, "step": 10345 }, { - "epoch": 0.57, - "learning_rate": 8.522405463934699e-06, - "loss": 0.344, + "epoch": 1.15, + "learning_rate": 8.520657485561973e-06, + "loss": 0.0753, "step": 10350 }, { - "epoch": 0.57, - "learning_rate": 8.516852684768728e-06, - "loss": 0.0907, + "epoch": 1.15, + "learning_rate": 8.515104398045314e-06, + "loss": 0.1312, "step": 10355 }, { - "epoch": 0.57, - "learning_rate": 8.511299905602755e-06, - "loss": 0.0824, + "epoch": 1.15, + "learning_rate": 8.509551310528655e-06, + "loss": 0.324, "step": 10360 }, { - "epoch": 0.58, - "learning_rate": 8.505747126436782e-06, - "loss": 0.1242, + "epoch": 1.15, + "learning_rate": 8.503998223011996e-06, + "loss": 0.069, "step": 10365 }, { - "epoch": 0.58, - "learning_rate": 8.500194347270809e-06, - "loss": 0.1467, + "epoch": 1.15, + "learning_rate": 8.498445135495335e-06, + "loss": 0.0876, "step": 10370 }, { - "epoch": 0.58, - "learning_rate": 8.494641568104837e-06, - "loss": 0.059, + "epoch": 1.15, + "learning_rate": 8.492892047978676e-06, + "loss": 0.0953, "step": 10375 }, { - "epoch": 0.58, - "learning_rate": 8.489088788938866e-06, - "loss": 0.115, + "epoch": 1.15, + "learning_rate": 8.487338960462017e-06, + "loss": 0.0984, "step": 10380 }, { - "epoch": 0.58, - "learning_rate": 8.483536009772893e-06, - "loss": 0.0565, + "epoch": 1.15, + "learning_rate": 8.481785872945358e-06, + "loss": 0.0535, "step": 10385 }, { - "epoch": 0.58, - "learning_rate": 8.47798323060692e-06, - "loss": 0.1551, + "epoch": 1.15, + "learning_rate": 8.476232785428698e-06, + "loss": 0.0803, "step": 10390 }, { - "epoch": 0.58, - "learning_rate": 8.472430451440947e-06, - "loss": 0.2112, + "epoch": 1.15, + "learning_rate": 8.47067969791204e-06, + "loss": 0.0545, "step": 10395 }, { - "epoch": 0.58, - "learning_rate": 8.466877672274975e-06, - "loss": 0.3947, + "epoch": 1.15, + "learning_rate": 8.46512661039538e-06, + "loss": 0.104, "step": 10400 }, { - "epoch": 0.58, - "learning_rate": 8.461324893109002e-06, - "loss": 0.1229, + "epoch": 1.15, + "learning_rate": 8.459573522878721e-06, + "loss": 0.1661, "step": 10405 }, { - "epoch": 0.58, - "learning_rate": 8.45577211394303e-06, - "loss": 0.1145, + "epoch": 1.16, + "learning_rate": 8.454020435362062e-06, + "loss": 0.2866, "step": 10410 }, { - "epoch": 0.58, - "learning_rate": 8.450219334777056e-06, - "loss": 0.1141, + "epoch": 1.16, + "learning_rate": 8.448467347845403e-06, + "loss": 0.0697, "step": 10415 }, { - "epoch": 0.58, - "learning_rate": 8.444666555611085e-06, - "loss": 0.1276, + "epoch": 1.16, + "learning_rate": 8.442914260328744e-06, + "loss": 0.0643, "step": 10420 }, { - "epoch": 0.58, - "learning_rate": 8.439113776445112e-06, - "loss": 0.0873, + "epoch": 1.16, + "learning_rate": 8.437361172812084e-06, + "loss": 0.0958, "step": 10425 }, { - "epoch": 0.58, - "learning_rate": 8.433560997279139e-06, - "loss": 0.0573, + "epoch": 1.16, + "learning_rate": 8.431808085295425e-06, + "loss": 0.0842, "step": 10430 }, { - "epoch": 0.58, - "learning_rate": 8.428008218113166e-06, - "loss": 0.1847, + "epoch": 1.16, + "learning_rate": 8.426254997778766e-06, + "loss": 0.0644, "step": 10435 }, { - "epoch": 0.58, - "learning_rate": 8.422455438947194e-06, - "loss": 0.1151, + "epoch": 1.16, + "learning_rate": 8.420701910262107e-06, + "loss": 0.0514, "step": 10440 }, { - "epoch": 0.58, - "learning_rate": 8.416902659781221e-06, - "loss": 0.105, + "epoch": 1.16, + "learning_rate": 8.415148822745446e-06, + "loss": 0.0719, "step": 10445 }, { - "epoch": 0.58, - "learning_rate": 8.411349880615248e-06, - "loss": 0.3515, + "epoch": 1.16, + "learning_rate": 8.409595735228789e-06, + "loss": 0.1048, "step": 10450 }, { - "epoch": 0.58, - "learning_rate": 8.405797101449275e-06, - "loss": 0.0914, + "epoch": 1.16, + "learning_rate": 8.404042647712128e-06, + "loss": 0.1167, "step": 10455 }, { - "epoch": 0.58, - "learning_rate": 8.400244322283304e-06, - "loss": 0.128, + "epoch": 1.16, + "learning_rate": 8.39848956019547e-06, + "loss": 0.2688, "step": 10460 }, { - "epoch": 0.58, - "learning_rate": 8.39469154311733e-06, - "loss": 0.1256, + "epoch": 1.16, + "learning_rate": 8.39293647267881e-06, + "loss": 0.0667, "step": 10465 }, { - "epoch": 0.58, - "learning_rate": 8.38913876395136e-06, - "loss": 0.1319, + "epoch": 1.16, + "learning_rate": 8.387383385162151e-06, + "loss": 0.0867, "step": 10470 }, { - "epoch": 0.58, - "learning_rate": 8.383585984785386e-06, - "loss": 0.0513, + "epoch": 1.16, + "learning_rate": 8.38183029764549e-06, + "loss": 0.0938, "step": 10475 }, { - "epoch": 0.58, - "learning_rate": 8.378033205619413e-06, - "loss": 0.076, + "epoch": 1.16, + "learning_rate": 8.376277210128832e-06, + "loss": 0.0971, "step": 10480 }, { - "epoch": 0.58, - "learning_rate": 8.37248042645344e-06, - "loss": 0.0709, + "epoch": 1.16, + "learning_rate": 8.370724122612173e-06, + "loss": 0.0371, "step": 10485 }, { - "epoch": 0.58, - "learning_rate": 8.366927647287469e-06, - "loss": 0.1105, + "epoch": 1.16, + "learning_rate": 8.365171035095514e-06, + "loss": 0.0456, "step": 10490 }, { - "epoch": 0.58, - "learning_rate": 8.361374868121496e-06, - "loss": 0.225, + "epoch": 1.16, + "learning_rate": 8.359617947578853e-06, + "loss": 0.0575, "step": 10495 }, { - "epoch": 0.58, - "learning_rate": 8.355822088955523e-06, - "loss": 0.3873, + "epoch": 1.17, + "learning_rate": 8.354064860062196e-06, + "loss": 0.1993, "step": 10500 }, - { - "epoch": 0.58, - "learning_rate": 8.35026930978955e-06, - "loss": 0.0825, + { + "epoch": 1.17, + "learning_rate": 8.348511772545536e-06, + "loss": 0.1699, "step": 10505 }, { - "epoch": 0.58, - "learning_rate": 8.344716530623578e-06, - "loss": 0.0911, + "epoch": 1.17, + "learning_rate": 8.342958685028877e-06, + "loss": 0.267, "step": 10510 }, { - "epoch": 0.58, - "learning_rate": 8.339163751457605e-06, - "loss": 0.1095, + "epoch": 1.17, + "learning_rate": 8.337405597512218e-06, + "loss": 0.0599, "step": 10515 }, { - "epoch": 0.58, - "learning_rate": 8.333610972291632e-06, - "loss": 0.1063, + "epoch": 1.17, + "learning_rate": 8.331852509995559e-06, + "loss": 0.0663, "step": 10520 }, { - "epoch": 0.58, - "learning_rate": 8.32805819312566e-06, - "loss": 0.0541, + "epoch": 1.17, + "learning_rate": 8.326299422478898e-06, + "loss": 0.0596, "step": 10525 }, { - "epoch": 0.58, - "learning_rate": 8.322505413959688e-06, - "loss": 0.1209, + "epoch": 1.17, + "learning_rate": 8.320746334962239e-06, + "loss": 0.0873, "step": 10530 }, { - "epoch": 0.58, - "learning_rate": 8.316952634793715e-06, - "loss": 0.1058, + "epoch": 1.17, + "learning_rate": 8.31519324744558e-06, + "loss": 0.0259, "step": 10535 }, { - "epoch": 0.58, - "learning_rate": 8.311399855627742e-06, - "loss": 0.1353, + "epoch": 1.17, + "learning_rate": 8.309640159928921e-06, + "loss": 0.0751, "step": 10540 }, { - "epoch": 0.59, - "learning_rate": 8.305847076461769e-06, - "loss": 0.2044, + "epoch": 1.17, + "learning_rate": 8.304087072412262e-06, + "loss": 0.057, "step": 10545 }, { - "epoch": 0.59, - "learning_rate": 8.300294297295797e-06, - "loss": 0.3861, + "epoch": 1.17, + "learning_rate": 8.298533984895603e-06, + "loss": 0.093, "step": 10550 }, { - "epoch": 0.59, - "learning_rate": 8.294741518129826e-06, - "loss": 0.1068, + "epoch": 1.17, + "learning_rate": 8.292980897378944e-06, + "loss": 0.0977, "step": 10555 }, { - "epoch": 0.59, - "learning_rate": 8.289188738963853e-06, - "loss": 0.0961, + "epoch": 1.17, + "learning_rate": 8.287427809862284e-06, + "loss": 0.239, "step": 10560 }, { - "epoch": 0.59, - "learning_rate": 8.28363595979788e-06, - "loss": 0.1157, + "epoch": 1.17, + "learning_rate": 8.281874722345625e-06, + "loss": 0.0785, "step": 10565 }, { - "epoch": 0.59, - "learning_rate": 8.278083180631907e-06, - "loss": 0.1265, + "epoch": 1.17, + "learning_rate": 8.276321634828966e-06, + "loss": 0.0693, "step": 10570 }, { - "epoch": 0.59, - "learning_rate": 8.272530401465936e-06, - "loss": 0.0532, + "epoch": 1.17, + "learning_rate": 8.270768547312307e-06, + "loss": 0.0745, "step": 10575 }, { - "epoch": 0.59, - "learning_rate": 8.266977622299962e-06, - "loss": 0.0219, + "epoch": 1.17, + "learning_rate": 8.265215459795646e-06, + "loss": 0.1049, "step": 10580 }, { - "epoch": 0.59, - "learning_rate": 8.26142484313399e-06, - "loss": 0.0579, + "epoch": 1.17, + "learning_rate": 8.259662372278987e-06, + "loss": 0.0331, "step": 10585 }, { - "epoch": 0.59, - "learning_rate": 8.255872063968016e-06, - "loss": 0.183, + "epoch": 1.18, + "learning_rate": 8.254109284762328e-06, + "loss": 0.0817, "step": 10590 }, { - "epoch": 0.59, - "learning_rate": 8.250319284802043e-06, - "loss": 0.1818, + "epoch": 1.18, + "learning_rate": 8.24855619724567e-06, + "loss": 0.0558, "step": 10595 }, { - "epoch": 0.59, - "learning_rate": 8.244766505636072e-06, - "loss": 0.3218, + "epoch": 1.18, + "learning_rate": 8.243003109729009e-06, + "loss": 0.0921, "step": 10600 }, { - "epoch": 0.59, - "learning_rate": 8.239213726470099e-06, - "loss": 0.0949, + "epoch": 1.18, + "learning_rate": 8.237450022212352e-06, + "loss": 0.228, "step": 10605 }, { - "epoch": 0.59, - "learning_rate": 8.233660947304126e-06, - "loss": 0.0958, + "epoch": 1.18, + "learning_rate": 8.231896934695691e-06, + "loss": 0.1868, "step": 10610 }, { - "epoch": 0.59, - "learning_rate": 8.228108168138153e-06, - "loss": 0.1433, + "epoch": 1.18, + "learning_rate": 8.226343847179032e-06, + "loss": 0.0731, "step": 10615 }, { - "epoch": 0.59, - "learning_rate": 8.222555388972181e-06, - "loss": 0.1182, + "epoch": 1.18, + "learning_rate": 8.220790759662373e-06, + "loss": 0.0744, "step": 10620 }, { - "epoch": 0.59, - "learning_rate": 8.217002609806208e-06, - "loss": 0.0669, + "epoch": 1.18, + "learning_rate": 8.215237672145714e-06, + "loss": 0.0719, "step": 10625 }, { - "epoch": 0.59, - "learning_rate": 8.211449830640235e-06, - "loss": 0.08, + "epoch": 1.18, + "learning_rate": 8.209684584629054e-06, + "loss": 0.0821, "step": 10630 }, { - "epoch": 0.59, - "learning_rate": 8.205897051474262e-06, - "loss": 0.1939, + "epoch": 1.18, + "learning_rate": 8.204131497112395e-06, + "loss": 0.0547, "step": 10635 }, { - "epoch": 0.59, - "learning_rate": 8.200344272308291e-06, - "loss": 0.1385, + "epoch": 1.18, + "learning_rate": 8.198578409595736e-06, + "loss": 0.0501, "step": 10640 }, { - "epoch": 0.59, - "learning_rate": 8.19479149314232e-06, - "loss": 0.1779, + "epoch": 1.18, + "learning_rate": 8.193025322079077e-06, + "loss": 0.054, "step": 10645 }, { - "epoch": 0.59, - "learning_rate": 8.189238713976347e-06, - "loss": 0.3574, + "epoch": 1.18, + "learning_rate": 8.187472234562416e-06, + "loss": 0.0934, "step": 10650 }, { - "epoch": 0.59, - "learning_rate": 8.183685934810373e-06, - "loss": 0.0859, + "epoch": 1.18, + "learning_rate": 8.181919147045759e-06, + "loss": 0.189, "step": 10655 }, { - "epoch": 0.59, - "learning_rate": 8.1781331556444e-06, - "loss": 0.1227, + "epoch": 1.18, + "learning_rate": 8.176366059529098e-06, + "loss": 0.1917, "step": 10660 }, { - "epoch": 0.59, - "learning_rate": 8.172580376478429e-06, - "loss": 0.117, + "epoch": 1.18, + "learning_rate": 8.17081297201244e-06, + "loss": 0.0897, "step": 10665 }, { - "epoch": 0.59, - "learning_rate": 8.167027597312456e-06, - "loss": 0.1041, + "epoch": 1.18, + "learning_rate": 8.16525988449578e-06, + "loss": 0.0714, "step": 10670 }, { - "epoch": 0.59, - "learning_rate": 8.161474818146483e-06, - "loss": 0.101, + "epoch": 1.18, + "learning_rate": 8.159706796979121e-06, + "loss": 0.0775, "step": 10675 }, { - "epoch": 0.59, - "learning_rate": 8.15592203898051e-06, - "loss": 0.0388, + "epoch": 1.19, + "learning_rate": 8.154153709462463e-06, + "loss": 0.098, "step": 10680 }, { - "epoch": 0.59, - "learning_rate": 8.150369259814539e-06, - "loss": 0.1609, + "epoch": 1.19, + "learning_rate": 8.148600621945802e-06, + "loss": 0.0518, "step": 10685 }, { - "epoch": 0.59, - "learning_rate": 8.144816480648565e-06, - "loss": 0.149, + "epoch": 1.19, + "learning_rate": 8.143047534429143e-06, + "loss": 0.0568, "step": 10690 }, { - "epoch": 0.59, - "learning_rate": 8.139263701482592e-06, - "loss": 0.1616, + "epoch": 1.19, + "learning_rate": 8.137494446912484e-06, + "loss": 0.069, "step": 10695 }, { - "epoch": 0.59, - "learning_rate": 8.13371092231662e-06, - "loss": 0.3101, + "epoch": 1.19, + "learning_rate": 8.131941359395825e-06, + "loss": 0.0887, "step": 10700 }, { - "epoch": 0.59, - "learning_rate": 8.128158143150648e-06, - "loss": 0.1189, + "epoch": 1.19, + "learning_rate": 8.126388271879166e-06, + "loss": 0.124, "step": 10705 }, { - "epoch": 0.59, - "learning_rate": 8.122605363984675e-06, - "loss": 0.0984, + "epoch": 1.19, + "learning_rate": 8.120835184362507e-06, + "loss": 0.249, "step": 10710 }, { - "epoch": 0.59, - "learning_rate": 8.117052584818702e-06, - "loss": 0.1162, + "epoch": 1.19, + "learning_rate": 8.115282096845847e-06, + "loss": 0.0698, "step": 10715 }, { - "epoch": 0.59, - "learning_rate": 8.111499805652729e-06, - "loss": 0.1305, + "epoch": 1.19, + "learning_rate": 8.109729009329188e-06, + "loss": 0.0587, "step": 10720 }, { - "epoch": 0.6, - "learning_rate": 8.105947026486758e-06, - "loss": 0.0656, + "epoch": 1.19, + "learning_rate": 8.104175921812529e-06, + "loss": 0.0617, "step": 10725 }, { - "epoch": 0.6, - "learning_rate": 8.100394247320784e-06, - "loss": 0.0204, + "epoch": 1.19, + "learning_rate": 8.09862283429587e-06, + "loss": 0.1027, "step": 10730 }, { - "epoch": 0.6, - "learning_rate": 8.094841468154813e-06, - "loss": 0.0582, + "epoch": 1.19, + "learning_rate": 8.093069746779209e-06, + "loss": 0.0704, "step": 10735 }, { - "epoch": 0.6, - "learning_rate": 8.08928868898884e-06, - "loss": 0.149, + "epoch": 1.19, + "learning_rate": 8.08751665926255e-06, + "loss": 0.0428, "step": 10740 }, { - "epoch": 0.6, - "learning_rate": 8.083735909822867e-06, - "loss": 0.2283, + "epoch": 1.19, + "learning_rate": 8.081963571745891e-06, + "loss": 0.0484, "step": 10745 }, { - "epoch": 0.6, - "learning_rate": 8.078183130656894e-06, - "loss": 0.4656, + "epoch": 1.19, + "learning_rate": 8.076410484229232e-06, + "loss": 0.1133, "step": 10750 }, { - "epoch": 0.6, - "learning_rate": 8.072630351490923e-06, - "loss": 0.085, + "epoch": 1.19, + "learning_rate": 8.070857396712572e-06, + "loss": 0.1798, "step": 10755 }, { - "epoch": 0.6, - "learning_rate": 8.06707757232495e-06, - "loss": 0.1054, + "epoch": 1.19, + "learning_rate": 8.065304309195914e-06, + "loss": 0.2342, "step": 10760 }, { - "epoch": 0.6, - "learning_rate": 8.061524793158977e-06, - "loss": 0.0999, + "epoch": 1.19, + "learning_rate": 8.059751221679254e-06, + "loss": 0.1101, "step": 10765 }, { - "epoch": 0.6, - "learning_rate": 8.055972013993003e-06, - "loss": 0.1272, + "epoch": 1.2, + "learning_rate": 8.054198134162595e-06, + "loss": 0.0627, "step": 10770 }, { - "epoch": 0.6, - "learning_rate": 8.050419234827032e-06, - "loss": 0.0877, + "epoch": 1.2, + "learning_rate": 8.048645046645936e-06, + "loss": 0.0895, "step": 10775 }, { - "epoch": 0.6, - "learning_rate": 8.044866455661059e-06, - "loss": 0.0801, + "epoch": 1.2, + "learning_rate": 8.043091959129277e-06, + "loss": 0.0515, "step": 10780 }, { - "epoch": 0.6, - "learning_rate": 8.039313676495086e-06, - "loss": 0.0634, + "epoch": 1.2, + "learning_rate": 8.037538871612616e-06, + "loss": 0.0282, "step": 10785 }, { - "epoch": 0.6, - "learning_rate": 8.033760897329113e-06, - "loss": 0.1141, + "epoch": 1.2, + "learning_rate": 8.031985784095957e-06, + "loss": 0.0678, "step": 10790 }, { - "epoch": 0.6, - "learning_rate": 8.028208118163142e-06, - "loss": 0.342, + "epoch": 1.2, + "learning_rate": 8.026432696579298e-06, + "loss": 0.0769, "step": 10795 }, { - "epoch": 0.6, - "learning_rate": 8.022655338997169e-06, - "loss": 0.344, + "epoch": 1.2, + "learning_rate": 8.02087960906264e-06, + "loss": 0.078, "step": 10800 }, { - "epoch": 0.6, - "learning_rate": 8.017102559831195e-06, - "loss": 0.0929, + "epoch": 1.2, + "learning_rate": 8.015326521545979e-06, + "loss": 0.2172, "step": 10805 }, { - "epoch": 0.6, - "learning_rate": 8.011549780665222e-06, - "loss": 0.1819, + "epoch": 1.2, + "learning_rate": 8.009773434029322e-06, + "loss": 0.3175, "step": 10810 }, { - "epoch": 0.6, - "learning_rate": 8.005997001499251e-06, - "loss": 0.097, + "epoch": 1.2, + "learning_rate": 8.004220346512663e-06, + "loss": 0.064, "step": 10815 }, { - "epoch": 0.6, - "learning_rate": 8.00044422233328e-06, - "loss": 0.137, + "epoch": 1.2, + "learning_rate": 7.998667258996002e-06, + "loss": 0.0797, "step": 10820 }, { - "epoch": 0.6, - "learning_rate": 7.994891443167307e-06, - "loss": 0.0613, + "epoch": 1.2, + "learning_rate": 7.993114171479343e-06, + "loss": 0.061, "step": 10825 }, { - "epoch": 0.6, - "learning_rate": 7.989338664001334e-06, - "loss": 0.0653, + "epoch": 1.2, + "learning_rate": 7.987561083962684e-06, + "loss": 0.0779, "step": 10830 }, { - "epoch": 0.6, - "learning_rate": 7.98378588483536e-06, - "loss": 0.0798, + "epoch": 1.2, + "learning_rate": 7.982007996446025e-06, + "loss": 0.0508, "step": 10835 }, { - "epoch": 0.6, - "learning_rate": 7.97823310566939e-06, - "loss": 0.1074, + "epoch": 1.2, + "learning_rate": 7.976454908929365e-06, + "loss": 0.0208, "step": 10840 }, { - "epoch": 0.6, - "learning_rate": 7.972680326503416e-06, - "loss": 0.1544, + "epoch": 1.2, + "learning_rate": 7.970901821412706e-06, + "loss": 0.069, "step": 10845 }, { - "epoch": 0.6, - "learning_rate": 7.967127547337443e-06, - "loss": 0.3255, + "epoch": 1.2, + "learning_rate": 7.965348733896047e-06, + "loss": 0.1362, "step": 10850 }, { - "epoch": 0.6, - "learning_rate": 7.96157476817147e-06, - "loss": 0.0773, + "epoch": 1.2, + "learning_rate": 7.959795646379388e-06, + "loss": 0.1576, "step": 10855 }, { - "epoch": 0.6, - "learning_rate": 7.956021989005499e-06, - "loss": 0.1246, + "epoch": 1.21, + "learning_rate": 7.954242558862729e-06, + "loss": 0.245, "step": 10860 }, { - "epoch": 0.6, - "learning_rate": 7.950469209839526e-06, - "loss": 0.0955, + "epoch": 1.21, + "learning_rate": 7.94868947134607e-06, + "loss": 0.0603, "step": 10865 }, { - "epoch": 0.6, - "learning_rate": 7.944916430673553e-06, - "loss": 0.1249, + "epoch": 1.21, + "learning_rate": 7.94313638382941e-06, + "loss": 0.0738, "step": 10870 }, { - "epoch": 0.6, - "learning_rate": 7.93936365150758e-06, - "loss": 0.1026, + "epoch": 1.21, + "learning_rate": 7.93758329631275e-06, + "loss": 0.0873, "step": 10875 }, { - "epoch": 0.6, - "learning_rate": 7.933810872341608e-06, - "loss": 0.108, + "epoch": 1.21, + "learning_rate": 7.932030208796091e-06, + "loss": 0.1042, "step": 10880 }, { - "epoch": 0.6, - "learning_rate": 7.928258093175635e-06, - "loss": 0.0604, + "epoch": 1.21, + "learning_rate": 7.926477121279433e-06, + "loss": 0.0486, "step": 10885 }, { - "epoch": 0.6, - "learning_rate": 7.922705314009662e-06, - "loss": 0.0789, + "epoch": 1.21, + "learning_rate": 7.920924033762772e-06, + "loss": 0.0378, "step": 10890 }, { - "epoch": 0.6, - "learning_rate": 7.917152534843689e-06, - "loss": 0.2054, + "epoch": 1.21, + "learning_rate": 7.915370946246113e-06, + "loss": 0.0678, "step": 10895 }, { - "epoch": 0.6, - "learning_rate": 7.911599755677718e-06, - "loss": 0.2302, + "epoch": 1.21, + "learning_rate": 7.909817858729454e-06, + "loss": 0.087, "step": 10900 }, { - "epoch": 0.61, - "learning_rate": 7.906046976511745e-06, - "loss": 0.1052, + "epoch": 1.21, + "learning_rate": 7.904264771212795e-06, + "loss": 0.1775, "step": 10905 }, { - "epoch": 0.61, - "learning_rate": 7.900494197345773e-06, - "loss": 0.1103, + "epoch": 1.21, + "learning_rate": 7.898711683696134e-06, + "loss": 0.2822, "step": 10910 }, { - "epoch": 0.61, - "learning_rate": 7.8949414181798e-06, - "loss": 0.0717, + "epoch": 1.21, + "learning_rate": 7.893158596179477e-06, + "loss": 0.0672, "step": 10915 }, { - "epoch": 0.61, - "learning_rate": 7.889388639013827e-06, - "loss": 0.0953, + "epoch": 1.21, + "learning_rate": 7.887605508662817e-06, + "loss": 0.0687, "step": 10920 }, { - "epoch": 0.61, - "learning_rate": 7.883835859847854e-06, - "loss": 0.1097, + "epoch": 1.21, + "learning_rate": 7.882052421146158e-06, + "loss": 0.0712, "step": 10925 }, { - "epoch": 0.61, - "learning_rate": 7.878283080681883e-06, - "loss": 0.0698, + "epoch": 1.21, + "learning_rate": 7.876499333629499e-06, + "loss": 0.0584, "step": 10930 }, { - "epoch": 0.61, - "learning_rate": 7.87273030151591e-06, - "loss": 0.0963, + "epoch": 1.21, + "learning_rate": 7.87094624611284e-06, + "loss": 0.0539, "step": 10935 }, { - "epoch": 0.61, - "learning_rate": 7.867177522349937e-06, - "loss": 0.1237, + "epoch": 1.21, + "learning_rate": 7.865393158596179e-06, + "loss": 0.0178, "step": 10940 }, { - "epoch": 0.61, - "learning_rate": 7.861624743183964e-06, - "loss": 0.1412, + "epoch": 1.21, + "learning_rate": 7.85984007107952e-06, + "loss": 0.0817, "step": 10945 }, { - "epoch": 0.61, - "learning_rate": 7.856071964017992e-06, - "loss": 0.506, + "epoch": 1.22, + "learning_rate": 7.854286983562863e-06, + "loss": 0.1031, "step": 10950 }, { - "epoch": 0.61, - "learning_rate": 7.85051918485202e-06, - "loss": 0.1595, + "epoch": 1.22, + "learning_rate": 7.848733896046202e-06, + "loss": 0.1493, "step": 10955 }, { - "epoch": 0.61, - "learning_rate": 7.844966405686046e-06, - "loss": 0.0943, + "epoch": 1.22, + "learning_rate": 7.843180808529543e-06, + "loss": 0.3111, "step": 10960 }, { - "epoch": 0.61, - "learning_rate": 7.839413626520073e-06, - "loss": 0.1561, + "epoch": 1.22, + "learning_rate": 7.837627721012884e-06, + "loss": 0.0735, "step": 10965 }, { - "epoch": 0.61, - "learning_rate": 7.833860847354102e-06, - "loss": 0.0876, + "epoch": 1.22, + "learning_rate": 7.832074633496226e-06, + "loss": 0.0746, "step": 10970 }, { - "epoch": 0.61, - "learning_rate": 7.828308068188129e-06, - "loss": 0.177, + "epoch": 1.22, + "learning_rate": 7.826521545979565e-06, + "loss": 0.0674, "step": 10975 }, { - "epoch": 0.61, - "learning_rate": 7.822755289022156e-06, - "loss": 0.1035, + "epoch": 1.22, + "learning_rate": 7.820968458462906e-06, + "loss": 0.0793, "step": 10980 }, { - "epoch": 0.61, - "learning_rate": 7.817202509856183e-06, - "loss": 0.0488, + "epoch": 1.22, + "learning_rate": 7.815415370946247e-06, + "loss": 0.0661, "step": 10985 }, { - "epoch": 0.61, - "learning_rate": 7.811649730690211e-06, - "loss": 0.1631, + "epoch": 1.22, + "learning_rate": 7.809862283429588e-06, + "loss": 0.0174, "step": 10990 }, { - "epoch": 0.61, - "learning_rate": 7.806096951524238e-06, - "loss": 0.1523, + "epoch": 1.22, + "learning_rate": 7.804309195912927e-06, + "loss": 0.0432, "step": 10995 }, { - "epoch": 0.61, - "learning_rate": 7.800544172358267e-06, - "loss": 0.2951, + "epoch": 1.22, + "learning_rate": 7.798756108396268e-06, + "loss": 0.0964, "step": 11000 }, { - "epoch": 0.61, - "learning_rate": 7.794991393192294e-06, - "loss": 0.0824, + "epoch": 1.22, + "learning_rate": 7.79320302087961e-06, + "loss": 0.1417, "step": 11005 }, { - "epoch": 0.61, - "learning_rate": 7.78943861402632e-06, - "loss": 0.1139, + "epoch": 1.22, + "learning_rate": 7.78764993336295e-06, + "loss": 0.2896, "step": 11010 }, { - "epoch": 0.61, - "learning_rate": 7.78388583486035e-06, - "loss": 0.1087, + "epoch": 1.22, + "learning_rate": 7.782096845846292e-06, + "loss": 0.0701, "step": 11015 }, { - "epoch": 0.61, - "learning_rate": 7.778333055694376e-06, - "loss": 0.1659, + "epoch": 1.22, + "learning_rate": 7.776543758329633e-06, + "loss": 0.0699, "step": 11020 }, { - "epoch": 0.61, - "learning_rate": 7.772780276528403e-06, - "loss": 0.0572, + "epoch": 1.22, + "learning_rate": 7.770990670812972e-06, + "loss": 0.0587, "step": 11025 }, { - "epoch": 0.61, - "learning_rate": 7.76722749736243e-06, - "loss": 0.0839, + "epoch": 1.22, + "learning_rate": 7.765437583296313e-06, + "loss": 0.0731, "step": 11030 }, { - "epoch": 0.61, - "learning_rate": 7.761674718196459e-06, - "loss": 0.1135, + "epoch": 1.22, + "learning_rate": 7.759884495779654e-06, + "loss": 0.054, "step": 11035 }, { - "epoch": 0.61, - "learning_rate": 7.756121939030486e-06, - "loss": 0.0887, + "epoch": 1.23, + "learning_rate": 7.754331408262995e-06, + "loss": 0.0428, "step": 11040 }, { - "epoch": 0.61, - "learning_rate": 7.750569159864513e-06, - "loss": 0.0826, + "epoch": 1.23, + "learning_rate": 7.748778320746335e-06, + "loss": 0.072, "step": 11045 }, { - "epoch": 0.61, - "learning_rate": 7.74501638069854e-06, - "loss": 0.2437, + "epoch": 1.23, + "learning_rate": 7.743225233229676e-06, + "loss": 0.1259, "step": 11050 }, { - "epoch": 0.61, - "learning_rate": 7.739463601532567e-06, - "loss": 0.1207, + "epoch": 1.23, + "learning_rate": 7.737672145713017e-06, + "loss": 0.077, "step": 11055 }, - { - "epoch": 0.61, - "learning_rate": 7.733910822366595e-06, - "loss": 0.1286, + { + "epoch": 1.23, + "learning_rate": 7.732119058196358e-06, + "loss": 0.1818, "step": 11060 }, { - "epoch": 0.61, - "learning_rate": 7.728358043200622e-06, - "loss": 0.1055, + "epoch": 1.23, + "learning_rate": 7.726565970679697e-06, + "loss": 0.0632, "step": 11065 }, { - "epoch": 0.61, - "learning_rate": 7.72280526403465e-06, - "loss": 0.1441, + "epoch": 1.23, + "learning_rate": 7.72101288316304e-06, + "loss": 0.0606, "step": 11070 }, { - "epoch": 0.61, - "learning_rate": 7.717252484868676e-06, - "loss": 0.1031, + "epoch": 1.23, + "learning_rate": 7.71545979564638e-06, + "loss": 0.0774, "step": 11075 }, { - "epoch": 0.61, - "learning_rate": 7.711699705702705e-06, - "loss": 0.0474, + "epoch": 1.23, + "learning_rate": 7.70990670812972e-06, + "loss": 0.0928, "step": 11080 }, { - "epoch": 0.62, - "learning_rate": 7.706146926536733e-06, - "loss": 0.1563, + "epoch": 1.23, + "learning_rate": 7.704353620613061e-06, + "loss": 0.0493, "step": 11085 }, { - "epoch": 0.62, - "learning_rate": 7.70059414737076e-06, - "loss": 0.1454, + "epoch": 1.23, + "learning_rate": 7.698800533096403e-06, + "loss": 0.074, "step": 11090 }, { - "epoch": 0.62, - "learning_rate": 7.695041368204787e-06, - "loss": 0.175, + "epoch": 1.23, + "learning_rate": 7.693247445579744e-06, + "loss": 0.0717, "step": 11095 }, { - "epoch": 0.62, - "learning_rate": 7.689488589038814e-06, - "loss": 0.2762, + "epoch": 1.23, + "learning_rate": 7.687694358063083e-06, + "loss": 0.0711, "step": 11100 }, { - "epoch": 0.62, - "learning_rate": 7.683935809872843e-06, - "loss": 0.0979, + "epoch": 1.23, + "learning_rate": 7.682141270546426e-06, + "loss": 0.1324, "step": 11105 }, { - "epoch": 0.62, - "learning_rate": 7.67838303070687e-06, - "loss": 0.1245, + "epoch": 1.23, + "learning_rate": 7.676588183029765e-06, + "loss": 0.2292, "step": 11110 }, { - "epoch": 0.62, - "learning_rate": 7.672830251540897e-06, - "loss": 0.0826, + "epoch": 1.23, + "learning_rate": 7.671035095513106e-06, + "loss": 0.0606, "step": 11115 }, { - "epoch": 0.62, - "learning_rate": 7.667277472374924e-06, - "loss": 0.1138, + "epoch": 1.23, + "learning_rate": 7.665482007996447e-06, + "loss": 0.0759, "step": 11120 }, { - "epoch": 0.62, - "learning_rate": 7.661724693208952e-06, - "loss": 0.1136, + "epoch": 1.23, + "learning_rate": 7.659928920479788e-06, + "loss": 0.0844, "step": 11125 }, { - "epoch": 0.62, - "learning_rate": 7.65617191404298e-06, - "loss": 0.0322, + "epoch": 1.24, + "learning_rate": 7.654375832963128e-06, + "loss": 0.0892, "step": 11130 }, { - "epoch": 0.62, - "learning_rate": 7.650619134877006e-06, - "loss": 0.0507, + "epoch": 1.24, + "learning_rate": 7.648822745446469e-06, + "loss": 0.066, "step": 11135 }, { - "epoch": 0.62, - "learning_rate": 7.645066355711033e-06, - "loss": 0.0818, + "epoch": 1.24, + "learning_rate": 7.64326965792981e-06, + "loss": 0.0444, "step": 11140 }, { - "epoch": 0.62, - "learning_rate": 7.639513576545062e-06, - "loss": 0.1246, + "epoch": 1.24, + "learning_rate": 7.63771657041315e-06, + "loss": 0.0964, "step": 11145 }, { - "epoch": 0.62, - "learning_rate": 7.633960797379089e-06, - "loss": 0.1736, + "epoch": 1.24, + "learning_rate": 7.63216348289649e-06, + "loss": 0.087, "step": 11150 }, { - "epoch": 0.62, - "learning_rate": 7.628408018213116e-06, - "loss": 0.0791, + "epoch": 1.24, + "learning_rate": 7.626610395379832e-06, + "loss": 0.2088, "step": 11155 }, { - "epoch": 0.62, - "learning_rate": 7.6228552390471436e-06, - "loss": 0.0942, + "epoch": 1.24, + "learning_rate": 7.621057307863172e-06, + "loss": 0.3376, "step": 11160 }, { - "epoch": 0.62, - "learning_rate": 7.6173024598811705e-06, - "loss": 0.0928, + "epoch": 1.24, + "learning_rate": 7.615504220346513e-06, + "loss": 0.058, "step": 11165 }, { - "epoch": 0.62, - "learning_rate": 7.611749680715198e-06, - "loss": 0.1168, + "epoch": 1.24, + "learning_rate": 7.609951132829854e-06, + "loss": 0.0705, "step": 11170 }, { - "epoch": 0.62, - "learning_rate": 7.606196901549226e-06, - "loss": 0.0987, + "epoch": 1.24, + "learning_rate": 7.604398045313195e-06, + "loss": 0.0804, "step": 11175 }, { - "epoch": 0.62, - "learning_rate": 7.600644122383254e-06, - "loss": 0.0914, + "epoch": 1.24, + "learning_rate": 7.598844957796535e-06, + "loss": 0.0868, "step": 11180 }, { - "epoch": 0.62, - "learning_rate": 7.595091343217281e-06, - "loss": 0.0763, + "epoch": 1.24, + "learning_rate": 7.593291870279876e-06, + "loss": 0.0629, "step": 11185 }, { - "epoch": 0.62, - "learning_rate": 7.589538564051309e-06, - "loss": 0.1642, + "epoch": 1.24, + "learning_rate": 7.587738782763216e-06, + "loss": 0.0387, "step": 11190 }, { - "epoch": 0.62, - "learning_rate": 7.585096340718531e-06, - "loss": 0.1932, + "epoch": 1.24, + "learning_rate": 7.582185695246558e-06, + "loss": 0.0682, "step": 11195 }, { - "epoch": 0.62, - "learning_rate": 7.5795435615525585e-06, - "loss": 0.3511, + "epoch": 1.24, + "learning_rate": 7.576632607729898e-06, + "loss": 0.0925, "step": 11200 }, { - "epoch": 0.62, - "learning_rate": 7.5739907823865855e-06, - "loss": 0.0912, + "epoch": 1.24, + "learning_rate": 7.571079520213239e-06, + "loss": 0.1252, "step": 11205 }, { - "epoch": 0.62, - "learning_rate": 7.568438003220613e-06, - "loss": 0.1163, + "epoch": 1.24, + "learning_rate": 7.5655264326965796e-06, + "loss": 0.2454, "step": 11210 }, { - "epoch": 0.62, - "learning_rate": 7.56288522405464e-06, - "loss": 0.1204, + "epoch": 1.24, + "learning_rate": 7.559973345179921e-06, + "loss": 0.0921, "step": 11215 }, { - "epoch": 0.62, - "learning_rate": 7.557332444888667e-06, - "loss": 0.1223, + "epoch": 1.25, + "learning_rate": 7.554420257663262e-06, + "loss": 0.0481, "step": 11220 }, { - "epoch": 0.62, - "learning_rate": 7.551779665722695e-06, - "loss": 0.0621, + "epoch": 1.25, + "learning_rate": 7.548867170146602e-06, + "loss": 0.0748, "step": 11225 }, { - "epoch": 0.62, - "learning_rate": 7.546226886556722e-06, - "loss": 0.0546, + "epoch": 1.25, + "learning_rate": 7.543314082629943e-06, + "loss": 0.0775, "step": 11230 }, { - "epoch": 0.62, - "learning_rate": 7.54067410739075e-06, - "loss": 0.145, + "epoch": 1.25, + "learning_rate": 7.537760995113283e-06, + "loss": 0.0666, "step": 11235 }, { - "epoch": 0.62, - "learning_rate": 7.535121328224777e-06, - "loss": 0.0662, + "epoch": 1.25, + "learning_rate": 7.532207907596625e-06, + "loss": 0.0388, "step": 11240 }, { - "epoch": 0.62, - "learning_rate": 7.5295685490588044e-06, - "loss": 0.212, + "epoch": 1.25, + "learning_rate": 7.526654820079965e-06, + "loss": 0.0551, "step": 11245 }, { - "epoch": 0.62, - "learning_rate": 7.524015769892831e-06, - "loss": 0.347, + "epoch": 1.25, + "learning_rate": 7.521101732563306e-06, + "loss": 0.087, "step": 11250 }, { - "epoch": 0.62, - "learning_rate": 7.518462990726859e-06, - "loss": 0.0826, + "epoch": 1.25, + "learning_rate": 7.5155486450466466e-06, + "loss": 0.1213, "step": 11255 }, { - "epoch": 0.62, - "learning_rate": 7.512910211560886e-06, - "loss": 0.1159, + "epoch": 1.25, + "learning_rate": 7.509995557529988e-06, + "loss": 0.2799, "step": 11260 }, { - "epoch": 0.63, - "learning_rate": 7.507357432394914e-06, - "loss": 0.106, + "epoch": 1.25, + "learning_rate": 7.504442470013328e-06, + "loss": 0.0806, "step": 11265 }, { - "epoch": 0.63, - "learning_rate": 7.501804653228941e-06, - "loss": 0.1171, + "epoch": 1.25, + "learning_rate": 7.498889382496669e-06, + "loss": 0.0787, "step": 11270 }, { - "epoch": 0.63, - "learning_rate": 7.496251874062969e-06, - "loss": 0.1081, + "epoch": 1.25, + "learning_rate": 7.493336294980009e-06, + "loss": 0.0837, "step": 11275 }, { - "epoch": 0.63, - "learning_rate": 7.490699094896997e-06, - "loss": 0.0664, + "epoch": 1.25, + "learning_rate": 7.48778320746335e-06, + "loss": 0.0846, "step": 11280 }, { - "epoch": 0.63, - "learning_rate": 7.485146315731024e-06, - "loss": 0.1205, + "epoch": 1.25, + "learning_rate": 7.48223011994669e-06, + "loss": 0.0462, "step": 11285 }, { - "epoch": 0.63, - "learning_rate": 7.479593536565052e-06, - "loss": 0.1319, + "epoch": 1.25, + "learning_rate": 7.476677032430032e-06, + "loss": 0.0767, "step": 11290 }, { - "epoch": 0.63, - "learning_rate": 7.474040757399079e-06, - "loss": 0.2215, + "epoch": 1.25, + "learning_rate": 7.4711239449133725e-06, + "loss": 0.0777, "step": 11295 }, { - "epoch": 0.63, - "learning_rate": 7.468487978233107e-06, - "loss": 0.2686, + "epoch": 1.25, + "learning_rate": 7.465570857396714e-06, + "loss": 0.099, "step": 11300 }, { - "epoch": 0.63, - "learning_rate": 7.462935199067134e-06, - "loss": 0.1167, + "epoch": 1.25, + "learning_rate": 7.460017769880054e-06, + "loss": 0.163, "step": 11305 }, { - "epoch": 0.63, - "learning_rate": 7.4573824199011616e-06, - "loss": 0.079, + "epoch": 1.26, + "learning_rate": 7.454464682363395e-06, + "loss": 0.2873, "step": 11310 }, { - "epoch": 0.63, - "learning_rate": 7.4518296407351885e-06, - "loss": 0.1015, + "epoch": 1.26, + "learning_rate": 7.448911594846735e-06, + "loss": 0.0684, "step": 11315 }, { - "epoch": 0.63, - "learning_rate": 7.446276861569216e-06, - "loss": 0.1152, + "epoch": 1.26, + "learning_rate": 7.443358507330076e-06, + "loss": 0.0565, "step": 11320 }, { - "epoch": 0.63, - "learning_rate": 7.440724082403243e-06, - "loss": 0.0996, + "epoch": 1.26, + "learning_rate": 7.437805419813416e-06, + "loss": 0.0902, "step": 11325 }, { - "epoch": 0.63, - "learning_rate": 7.435171303237271e-06, - "loss": 0.0366, + "epoch": 1.26, + "learning_rate": 7.432252332296757e-06, + "loss": 0.0835, "step": 11330 }, { - "epoch": 0.63, - "learning_rate": 7.429618524071298e-06, - "loss": 0.0952, + "epoch": 1.26, + "learning_rate": 7.426699244780098e-06, + "loss": 0.0621, "step": 11335 }, { - "epoch": 0.63, - "learning_rate": 7.424065744905326e-06, - "loss": 0.1797, + "epoch": 1.26, + "learning_rate": 7.421146157263439e-06, + "loss": 0.0421, "step": 11340 }, { - "epoch": 0.63, - "learning_rate": 7.418512965739353e-06, - "loss": 0.2037, + "epoch": 1.26, + "learning_rate": 7.415593069746779e-06, + "loss": 0.0957, "step": 11345 }, { - "epoch": 0.63, - "learning_rate": 7.4129601865733805e-06, - "loss": 0.348, + "epoch": 1.26, + "learning_rate": 7.410039982230121e-06, + "loss": 0.0845, "step": 11350 }, { - "epoch": 0.63, - "learning_rate": 7.4074074074074075e-06, - "loss": 0.0685, + "epoch": 1.26, + "learning_rate": 7.404486894713462e-06, + "loss": 0.1433, "step": 11355 }, { - "epoch": 0.63, - "learning_rate": 7.401854628241435e-06, - "loss": 0.113, + "epoch": 1.26, + "learning_rate": 7.398933807196802e-06, + "loss": 0.1906, "step": 11360 }, { - "epoch": 0.63, - "learning_rate": 7.396301849075462e-06, - "loss": 0.087, + "epoch": 1.26, + "learning_rate": 7.393380719680143e-06, + "loss": 0.0645, "step": 11365 }, { - "epoch": 0.63, - "learning_rate": 7.390749069909491e-06, - "loss": 0.1117, + "epoch": 1.26, + "learning_rate": 7.387827632163483e-06, + "loss": 0.0532, "step": 11370 }, { - "epoch": 0.63, - "learning_rate": 7.385196290743518e-06, - "loss": 0.0911, + "epoch": 1.26, + "learning_rate": 7.3822745446468244e-06, + "loss": 0.0664, "step": 11375 }, { - "epoch": 0.63, - "learning_rate": 7.379643511577546e-06, - "loss": 0.0817, + "epoch": 1.26, + "learning_rate": 7.376721457130165e-06, + "loss": 0.0747, "step": 11380 }, { - "epoch": 0.63, - "learning_rate": 7.3740907324115726e-06, - "loss": 0.075, + "epoch": 1.26, + "learning_rate": 7.371168369613506e-06, + "loss": 0.0381, "step": 11385 }, { - "epoch": 0.63, - "learning_rate": 7.3685379532456e-06, - "loss": 0.1192, + "epoch": 1.26, + "learning_rate": 7.365615282096846e-06, + "loss": 0.0541, "step": 11390 }, { - "epoch": 0.63, - "learning_rate": 7.362985174079627e-06, - "loss": 0.1953, + "epoch": 1.26, + "learning_rate": 7.360062194580188e-06, + "loss": 0.0625, "step": 11395 }, { - "epoch": 0.63, - "learning_rate": 7.357432394913655e-06, - "loss": 0.3567, + "epoch": 1.27, + "learning_rate": 7.354509107063528e-06, + "loss": 0.0858, "step": 11400 }, { - "epoch": 0.63, - "learning_rate": 7.351879615747682e-06, - "loss": 0.095, + "epoch": 1.27, + "learning_rate": 7.348956019546869e-06, + "loss": 0.1765, "step": 11405 }, { - "epoch": 0.63, - "learning_rate": 7.34632683658171e-06, - "loss": 0.0824, + "epoch": 1.27, + "learning_rate": 7.343402932030209e-06, + "loss": 0.3694, "step": 11410 }, { - "epoch": 0.63, - "learning_rate": 7.340774057415737e-06, - "loss": 0.0995, + "epoch": 1.27, + "learning_rate": 7.33784984451355e-06, + "loss": 0.0708, "step": 11415 }, { - "epoch": 0.63, - "learning_rate": 7.335221278249765e-06, - "loss": 0.098, + "epoch": 1.27, + "learning_rate": 7.332296756996891e-06, + "loss": 0.0632, "step": 11420 }, { - "epoch": 0.63, - "learning_rate": 7.3296684990837915e-06, - "loss": 0.0849, + "epoch": 1.27, + "learning_rate": 7.326743669480232e-06, + "loss": 0.0825, "step": 11425 }, { - "epoch": 0.63, - "learning_rate": 7.324115719917819e-06, - "loss": 0.1014, + "epoch": 1.27, + "learning_rate": 7.321190581963572e-06, + "loss": 0.0794, "step": 11430 }, { - "epoch": 0.63, - "learning_rate": 7.318562940751846e-06, - "loss": 0.1095, + "epoch": 1.27, + "learning_rate": 7.315637494446913e-06, + "loss": 0.0507, "step": 11435 }, { - "epoch": 0.63, - "learning_rate": 7.313010161585874e-06, - "loss": 0.1433, + "epoch": 1.27, + "learning_rate": 7.310084406930253e-06, + "loss": 0.072, "step": 11440 }, { - "epoch": 0.64, - "learning_rate": 7.307457382419901e-06, - "loss": 0.1593, + "epoch": 1.27, + "learning_rate": 7.304531319413595e-06, + "loss": 0.049, "step": 11445 }, { - "epoch": 0.64, - "learning_rate": 7.301904603253929e-06, - "loss": 0.3003, + "epoch": 1.27, + "learning_rate": 7.298978231896935e-06, + "loss": 0.0754, "step": 11450 }, { - "epoch": 0.64, - "learning_rate": 7.2963518240879575e-06, - "loss": 0.0901, + "epoch": 1.27, + "learning_rate": 7.293425144380276e-06, + "loss": 0.139, "step": 11455 }, { - "epoch": 0.64, - "learning_rate": 7.290799044921984e-06, - "loss": 0.1192, + "epoch": 1.27, + "learning_rate": 7.2878720568636166e-06, + "loss": 0.21, "step": 11460 }, { - "epoch": 0.64, - "learning_rate": 7.285246265756012e-06, - "loss": 0.1263, + "epoch": 1.27, + "learning_rate": 7.282318969346958e-06, + "loss": 0.0622, "step": 11465 }, { - "epoch": 0.64, - "learning_rate": 7.279693486590039e-06, - "loss": 0.116, + "epoch": 1.27, + "learning_rate": 7.276765881830298e-06, + "loss": 0.0679, "step": 11470 }, { - "epoch": 0.64, - "learning_rate": 7.274140707424067e-06, - "loss": 0.0446, + "epoch": 1.27, + "learning_rate": 7.271212794313639e-06, + "loss": 0.0903, "step": 11475 }, { - "epoch": 0.64, - "learning_rate": 7.268587928258094e-06, - "loss": 0.0632, + "epoch": 1.27, + "learning_rate": 7.265659706796979e-06, + "loss": 0.0885, "step": 11480 }, { - "epoch": 0.64, - "learning_rate": 7.263035149092122e-06, - "loss": 0.1635, + "epoch": 1.27, + "learning_rate": 7.26010661928032e-06, + "loss": 0.0284, "step": 11485 }, { - "epoch": 0.64, - "learning_rate": 7.257482369926149e-06, - "loss": 0.1321, + "epoch": 1.28, + "learning_rate": 7.254553531763662e-06, + "loss": 0.0431, "step": 11490 }, { - "epoch": 0.64, - "learning_rate": 7.2519295907601764e-06, - "loss": 0.1827, + "epoch": 1.28, + "learning_rate": 7.2490004442470015e-06, + "loss": 0.0551, "step": 11495 }, { - "epoch": 0.64, - "learning_rate": 7.246376811594203e-06, - "loss": 0.4158, + "epoch": 1.28, + "learning_rate": 7.243447356730343e-06, + "loss": 0.1224, "step": 11500 }, { - "epoch": 0.64, - "learning_rate": 7.240824032428231e-06, - "loss": 0.0796, + "epoch": 1.28, + "learning_rate": 7.237894269213684e-06, + "loss": 0.1565, "step": 11505 }, { - "epoch": 0.64, - "learning_rate": 7.235271253262258e-06, - "loss": 0.127, + "epoch": 1.28, + "learning_rate": 7.232341181697025e-06, + "loss": 0.24, "step": 11510 }, { - "epoch": 0.64, - "learning_rate": 7.229718474096286e-06, - "loss": 0.067, + "epoch": 1.28, + "learning_rate": 7.226788094180365e-06, + "loss": 0.077, "step": 11515 }, { - "epoch": 0.64, - "learning_rate": 7.224165694930313e-06, - "loss": 0.1837, + "epoch": 1.28, + "learning_rate": 7.221235006663706e-06, + "loss": 0.0688, "step": 11520 }, { - "epoch": 0.64, - "learning_rate": 7.218612915764341e-06, - "loss": 0.0961, + "epoch": 1.28, + "learning_rate": 7.215681919147046e-06, + "loss": 0.0758, "step": 11525 }, { - "epoch": 0.64, - "learning_rate": 7.213060136598368e-06, - "loss": 0.0739, + "epoch": 1.28, + "learning_rate": 7.210128831630387e-06, + "loss": 0.1004, "step": 11530 }, { - "epoch": 0.64, - "learning_rate": 7.207507357432395e-06, - "loss": 0.1084, + "epoch": 1.28, + "learning_rate": 7.2045757441137274e-06, + "loss": 0.0554, "step": 11535 }, { - "epoch": 0.64, - "learning_rate": 7.201954578266422e-06, - "loss": 0.1231, + "epoch": 1.28, + "learning_rate": 7.1990226565970685e-06, + "loss": 0.0266, "step": 11540 }, { - "epoch": 0.64, - "learning_rate": 7.196401799100451e-06, - "loss": 0.1969, + "epoch": 1.28, + "learning_rate": 7.193469569080409e-06, + "loss": 0.0425, "step": 11545 }, { - "epoch": 0.64, - "learning_rate": 7.190849019934478e-06, - "loss": 0.3598, + "epoch": 1.28, + "learning_rate": 7.187916481563751e-06, + "loss": 0.0686, "step": 11550 }, { - "epoch": 0.64, - "learning_rate": 7.185296240768506e-06, - "loss": 0.1066, + "epoch": 1.28, + "learning_rate": 7.182363394047091e-06, + "loss": 0.1635, "step": 11555 }, { - "epoch": 0.64, - "learning_rate": 7.179743461602533e-06, - "loss": 0.0968, + "epoch": 1.28, + "learning_rate": 7.176810306530432e-06, + "loss": 0.2836, "step": 11560 }, { - "epoch": 0.64, - "learning_rate": 7.1741906824365605e-06, - "loss": 0.1114, + "epoch": 1.28, + "learning_rate": 7.171257219013772e-06, + "loss": 0.0675, "step": 11565 }, { - "epoch": 0.64, - "learning_rate": 7.1686379032705875e-06, - "loss": 0.1079, + "epoch": 1.28, + "learning_rate": 7.165704131497113e-06, + "loss": 0.0556, "step": 11570 }, { - "epoch": 0.64, - "learning_rate": 7.163085124104615e-06, - "loss": 0.1224, + "epoch": 1.28, + "learning_rate": 7.160151043980453e-06, + "loss": 0.0896, "step": 11575 }, { - "epoch": 0.64, - "learning_rate": 7.157532344938642e-06, - "loss": 0.0932, + "epoch": 1.29, + "learning_rate": 7.1545979564637944e-06, + "loss": 0.1134, "step": 11580 }, { - "epoch": 0.64, - "learning_rate": 7.15197956577267e-06, - "loss": 0.0651, + "epoch": 1.29, + "learning_rate": 7.149044868947135e-06, + "loss": 0.0585, "step": 11585 }, { - "epoch": 0.64, - "learning_rate": 7.146426786606697e-06, - "loss": 0.1661, + "epoch": 1.29, + "learning_rate": 7.143491781430476e-06, + "loss": 0.0608, "step": 11590 }, { - "epoch": 0.64, - "learning_rate": 7.140874007440725e-06, - "loss": 0.2016, + "epoch": 1.29, + "learning_rate": 7.137938693913816e-06, + "loss": 0.0798, "step": 11595 }, { - "epoch": 0.64, - "learning_rate": 7.135321228274752e-06, - "loss": 0.4077, + "epoch": 1.29, + "learning_rate": 7.132385606397158e-06, + "loss": 0.0943, "step": 11600 }, { - "epoch": 0.64, - "learning_rate": 7.1297684491087795e-06, - "loss": 0.1171, + "epoch": 1.29, + "learning_rate": 7.126832518880498e-06, + "loss": 0.1184, "step": 11605 }, { - "epoch": 0.64, - "learning_rate": 7.1242156699428064e-06, - "loss": 0.1337, + "epoch": 1.29, + "learning_rate": 7.121279431363839e-06, + "loss": 0.1871, "step": 11610 }, { - "epoch": 0.64, - "learning_rate": 7.118662890776834e-06, - "loss": 0.1225, + "epoch": 1.29, + "learning_rate": 7.115726343847179e-06, + "loss": 0.0624, "step": 11615 }, { - "epoch": 0.64, - "learning_rate": 7.113110111610861e-06, - "loss": 0.1019, + "epoch": 1.29, + "learning_rate": 7.11017325633052e-06, + "loss": 0.0699, "step": 11620 }, { - "epoch": 0.65, - "learning_rate": 7.107557332444889e-06, - "loss": 0.1014, + "epoch": 1.29, + "learning_rate": 7.1046201688138615e-06, + "loss": 0.0656, "step": 11625 }, { - "epoch": 0.65, - "learning_rate": 7.102004553278916e-06, - "loss": 0.0864, + "epoch": 1.29, + "learning_rate": 7.099067081297202e-06, + "loss": 0.0663, "step": 11630 }, { - "epoch": 0.65, - "learning_rate": 7.0964517741129446e-06, - "loss": 0.0687, + "epoch": 1.29, + "learning_rate": 7.093513993780543e-06, + "loss": 0.0486, "step": 11635 }, { - "epoch": 0.65, - "learning_rate": 7.090898994946972e-06, - "loss": 0.0992, + "epoch": 1.29, + "learning_rate": 7.087960906263883e-06, + "loss": 0.052, "step": 11640 }, { - "epoch": 0.65, - "learning_rate": 7.085346215780999e-06, - "loss": 0.1179, + "epoch": 1.29, + "learning_rate": 7.082407818747225e-06, + "loss": 0.1055, "step": 11645 }, { - "epoch": 0.65, - "learning_rate": 7.079793436615027e-06, - "loss": 0.3071, + "epoch": 1.29, + "learning_rate": 7.076854731230564e-06, + "loss": 0.0685, "step": 11650 }, { - "epoch": 0.65, - "learning_rate": 7.074240657449054e-06, - "loss": 0.1071, + "epoch": 1.29, + "learning_rate": 7.071301643713906e-06, + "loss": 0.114, "step": 11655 }, { - "epoch": 0.65, - "learning_rate": 7.068687878283082e-06, - "loss": 0.0792, + "epoch": 1.29, + "learning_rate": 7.065748556197246e-06, + "loss": 0.2795, "step": 11660 }, { - "epoch": 0.65, - "learning_rate": 7.063135099117109e-06, - "loss": 0.1314, + "epoch": 1.29, + "learning_rate": 7.060195468680587e-06, + "loss": 0.1081, "step": 11665 }, { - "epoch": 0.65, - "learning_rate": 7.057582319951137e-06, - "loss": 0.1245, + "epoch": 1.3, + "learning_rate": 7.054642381163928e-06, + "loss": 0.056, "step": 11670 }, { - "epoch": 0.65, - "learning_rate": 7.0520295407851635e-06, - "loss": 0.1376, + "epoch": 1.3, + "learning_rate": 7.049089293647269e-06, + "loss": 0.0806, "step": 11675 }, { - "epoch": 0.65, - "learning_rate": 7.046476761619191e-06, - "loss": 0.086, + "epoch": 1.3, + "learning_rate": 7.043536206130609e-06, + "loss": 0.0957, "step": 11680 }, { - "epoch": 0.65, - "learning_rate": 7.040923982453218e-06, - "loss": 0.0835, + "epoch": 1.3, + "learning_rate": 7.03798311861395e-06, + "loss": 0.0456, "step": 11685 }, { - "epoch": 0.65, - "learning_rate": 7.035371203287245e-06, - "loss": 0.1404, + "epoch": 1.3, + "learning_rate": 7.03243003109729e-06, + "loss": 0.0738, "step": 11690 }, { - "epoch": 0.65, - "learning_rate": 7.029818424121273e-06, - "loss": 0.1928, + "epoch": 1.3, + "learning_rate": 7.026876943580631e-06, + "loss": 0.082, "step": 11695 }, { - "epoch": 0.65, - "learning_rate": 7.0242656449553e-06, - "loss": 0.2931, + "epoch": 1.3, + "learning_rate": 7.0213238560639715e-06, + "loss": 0.0751, "step": 11700 }, { - "epoch": 0.65, - "learning_rate": 7.018712865789328e-06, - "loss": 0.08, + "epoch": 1.3, + "learning_rate": 7.015770768547313e-06, + "loss": 0.1192, "step": 11705 }, { - "epoch": 0.65, - "learning_rate": 7.013160086623355e-06, - "loss": 0.0948, + "epoch": 1.3, + "learning_rate": 7.010217681030654e-06, + "loss": 0.2127, "step": 11710 }, { - "epoch": 0.65, - "learning_rate": 7.0076073074573825e-06, - "loss": 0.1024, + "epoch": 1.3, + "learning_rate": 7.004664593513995e-06, + "loss": 0.042, "step": 11715 }, { - "epoch": 0.65, - "learning_rate": 7.002054528291411e-06, - "loss": 0.1471, + "epoch": 1.3, + "learning_rate": 6.999111505997335e-06, + "loss": 0.0678, "step": 11720 }, { - "epoch": 0.65, - "learning_rate": 6.996501749125438e-06, - "loss": 0.1176, + "epoch": 1.3, + "learning_rate": 6.993558418480676e-06, + "loss": 0.0663, "step": 11725 }, { - "epoch": 0.65, - "learning_rate": 6.990948969959466e-06, - "loss": 0.0622, + "epoch": 1.3, + "learning_rate": 6.988005330964016e-06, + "loss": 0.0781, "step": 11730 }, { - "epoch": 0.65, - "learning_rate": 6.985396190793493e-06, - "loss": 0.084, + "epoch": 1.3, + "learning_rate": 6.982452243447357e-06, + "loss": 0.077, "step": 11735 }, { - "epoch": 0.65, - "learning_rate": 6.979843411627521e-06, - "loss": 0.1048, + "epoch": 1.3, + "learning_rate": 6.9768991559306974e-06, + "loss": 0.0657, "step": 11740 }, { - "epoch": 0.65, - "learning_rate": 6.974290632461548e-06, - "loss": 0.1823, + "epoch": 1.3, + "learning_rate": 6.9713460684140385e-06, + "loss": 0.0659, "step": 11745 }, { - "epoch": 0.65, - "learning_rate": 6.968737853295575e-06, - "loss": 0.208, + "epoch": 1.3, + "learning_rate": 6.965792980897379e-06, + "loss": 0.0878, "step": 11750 }, { - "epoch": 0.65, - "learning_rate": 6.963185074129602e-06, - "loss": 0.1277, + "epoch": 1.3, + "learning_rate": 6.960239893380721e-06, + "loss": 0.0875, "step": 11755 }, { - "epoch": 0.65, - "learning_rate": 6.95763229496363e-06, - "loss": 0.1041, + "epoch": 1.31, + "learning_rate": 6.954686805864062e-06, + "loss": 0.1752, "step": 11760 }, { - "epoch": 0.65, - "learning_rate": 6.952079515797657e-06, - "loss": 0.0897, + "epoch": 1.31, + "learning_rate": 6.949133718347402e-06, + "loss": 0.0628, "step": 11765 }, { - "epoch": 0.65, - "learning_rate": 6.946526736631685e-06, - "loss": 0.155, + "epoch": 1.31, + "learning_rate": 6.943580630830743e-06, + "loss": 0.0576, "step": 11770 }, { - "epoch": 0.65, - "learning_rate": 6.940973957465712e-06, - "loss": 0.0415, + "epoch": 1.31, + "learning_rate": 6.938027543314083e-06, + "loss": 0.0735, "step": 11775 }, { - "epoch": 0.65, - "learning_rate": 6.93542117829974e-06, - "loss": 0.0646, + "epoch": 1.31, + "learning_rate": 6.932474455797424e-06, + "loss": 0.0748, "step": 11780 }, { - "epoch": 0.65, - "learning_rate": 6.9298683991337666e-06, - "loss": 0.1393, + "epoch": 1.31, + "learning_rate": 6.9269213682807644e-06, + "loss": 0.046, "step": 11785 }, { - "epoch": 0.65, - "learning_rate": 6.924315619967794e-06, - "loss": 0.1032, + "epoch": 1.31, + "learning_rate": 6.9213682807641055e-06, + "loss": 0.0389, "step": 11790 }, { - "epoch": 0.65, - "learning_rate": 6.918762840801821e-06, - "loss": 0.1132, + "epoch": 1.31, + "learning_rate": 6.915815193247446e-06, + "loss": 0.0605, "step": 11795 }, { - "epoch": 0.65, - "learning_rate": 6.913210061635849e-06, - "loss": 0.3368, + "epoch": 1.31, + "learning_rate": 6.910262105730788e-06, + "loss": 0.0961, "step": 11800 }, { - "epoch": 0.66, - "learning_rate": 6.907657282469876e-06, - "loss": 0.0834, + "epoch": 1.31, + "learning_rate": 6.904709018214127e-06, + "loss": 0.1166, "step": 11805 }, { - "epoch": 0.66, - "learning_rate": 6.902104503303905e-06, - "loss": 0.1003, + "epoch": 1.31, + "learning_rate": 6.899155930697469e-06, + "loss": 0.258, "step": 11810 }, { - "epoch": 0.66, - "learning_rate": 6.896551724137932e-06, - "loss": 0.1452, + "epoch": 1.31, + "learning_rate": 6.893602843180809e-06, + "loss": 0.0542, "step": 11815 }, { - "epoch": 0.66, - "learning_rate": 6.8909989449719595e-06, - "loss": 0.1114, + "epoch": 1.31, + "learning_rate": 6.88804975566415e-06, + "loss": 0.0674, "step": 11820 }, { - "epoch": 0.66, - "learning_rate": 6.885446165805986e-06, - "loss": 0.0588, + "epoch": 1.31, + "learning_rate": 6.88249666814749e-06, + "loss": 0.0864, "step": 11825 }, { - "epoch": 0.66, - "learning_rate": 6.879893386640014e-06, - "loss": 0.1104, + "epoch": 1.31, + "learning_rate": 6.8769435806308315e-06, + "loss": 0.0839, "step": 11830 }, { - "epoch": 0.66, - "learning_rate": 6.874340607474041e-06, - "loss": 0.1159, + "epoch": 1.31, + "learning_rate": 6.871390493114172e-06, + "loss": 0.0785, "step": 11835 }, { - "epoch": 0.66, - "learning_rate": 6.868787828308069e-06, - "loss": 0.1358, + "epoch": 1.31, + "learning_rate": 6.865837405597513e-06, + "loss": 0.0333, "step": 11840 }, { - "epoch": 0.66, - "learning_rate": 6.863235049142096e-06, - "loss": 0.3362, + "epoch": 1.31, + "learning_rate": 6.860284318080853e-06, + "loss": 0.0789, "step": 11845 }, { - "epoch": 0.66, - "learning_rate": 6.857682269976124e-06, - "loss": 0.3762, + "epoch": 1.32, + "learning_rate": 6.854731230564194e-06, + "loss": 0.0888, "step": 11850 }, { - "epoch": 0.66, - "learning_rate": 6.852129490810151e-06, - "loss": 0.1069, + "epoch": 1.32, + "learning_rate": 6.849178143047534e-06, + "loss": 0.1334, "step": 11855 }, { - "epoch": 0.66, - "learning_rate": 6.8465767116441784e-06, - "loss": 0.1181, + "epoch": 1.32, + "learning_rate": 6.843625055530876e-06, + "loss": 0.2851, "step": 11860 }, { - "epoch": 0.66, - "learning_rate": 6.841023932478205e-06, - "loss": 0.1181, + "epoch": 1.32, + "learning_rate": 6.838071968014216e-06, + "loss": 0.0718, "step": 11865 }, { - "epoch": 0.66, - "learning_rate": 6.835471153312233e-06, - "loss": 0.1604, + "epoch": 1.32, + "learning_rate": 6.832518880497557e-06, + "loss": 0.0622, "step": 11870 }, { - "epoch": 0.66, - "learning_rate": 6.82991837414626e-06, - "loss": 0.078, + "epoch": 1.32, + "learning_rate": 6.826965792980898e-06, + "loss": 0.0734, "step": 11875 }, { - "epoch": 0.66, - "learning_rate": 6.824365594980288e-06, - "loss": 0.0398, + "epoch": 1.32, + "learning_rate": 6.821412705464239e-06, + "loss": 0.0712, "step": 11880 }, { - "epoch": 0.66, - "learning_rate": 6.818812815814315e-06, - "loss": 0.0867, + "epoch": 1.32, + "learning_rate": 6.815859617947579e-06, + "loss": 0.0618, "step": 11885 }, { - "epoch": 0.66, - "learning_rate": 6.813260036648343e-06, - "loss": 0.1419, + "epoch": 1.32, + "learning_rate": 6.81030653043092e-06, + "loss": 0.0463, "step": 11890 }, { - "epoch": 0.66, - "learning_rate": 6.80770725748237e-06, - "loss": 0.2253, + "epoch": 1.32, + "learning_rate": 6.804753442914261e-06, + "loss": 0.0656, "step": 11895 }, { - "epoch": 0.66, - "learning_rate": 6.802154478316398e-06, - "loss": 0.3917, + "epoch": 1.32, + "learning_rate": 6.799200355397601e-06, + "loss": 0.1304, "step": 11900 }, { - "epoch": 0.66, - "learning_rate": 6.796601699150426e-06, - "loss": 0.0953, + "epoch": 1.32, + "learning_rate": 6.793647267880943e-06, + "loss": 0.1738, "step": 11905 }, { - "epoch": 0.66, - "learning_rate": 6.791048919984453e-06, - "loss": 0.1445, + "epoch": 1.32, + "learning_rate": 6.788094180364283e-06, + "loss": 0.3324, "step": 11910 }, { - "epoch": 0.66, - "learning_rate": 6.785496140818481e-06, - "loss": 0.0742, + "epoch": 1.32, + "learning_rate": 6.7825410928476244e-06, + "loss": 0.0565, "step": 11915 }, { - "epoch": 0.66, - "learning_rate": 6.779943361652508e-06, - "loss": 0.1148, + "epoch": 1.32, + "learning_rate": 6.776988005330965e-06, + "loss": 0.0709, "step": 11920 }, { - "epoch": 0.66, - "learning_rate": 6.7743905824865355e-06, - "loss": 0.1171, + "epoch": 1.32, + "learning_rate": 6.771434917814306e-06, + "loss": 0.0738, "step": 11925 }, { - "epoch": 0.66, - "learning_rate": 6.7688378033205625e-06, - "loss": 0.0467, + "epoch": 1.32, + "learning_rate": 6.765881830297646e-06, + "loss": 0.0898, "step": 11930 }, { - "epoch": 0.66, - "learning_rate": 6.76328502415459e-06, - "loss": 0.1578, + "epoch": 1.32, + "learning_rate": 6.760328742780987e-06, + "loss": 0.0875, "step": 11935 }, { - "epoch": 0.66, - "learning_rate": 6.757732244988617e-06, - "loss": 0.1481, + "epoch": 1.33, + "learning_rate": 6.754775655264327e-06, + "loss": 0.0233, "step": 11940 }, { - "epoch": 0.66, - "learning_rate": 6.752179465822645e-06, - "loss": 0.2591, + "epoch": 1.33, + "learning_rate": 6.749222567747668e-06, + "loss": 0.0473, "step": 11945 }, { - "epoch": 0.66, - "learning_rate": 6.746626686656672e-06, - "loss": 0.4473, + "epoch": 1.33, + "learning_rate": 6.7436694802310085e-06, + "loss": 0.1188, "step": 11950 }, { - "epoch": 0.66, - "learning_rate": 6.7410739074907e-06, - "loss": 0.0924, + "epoch": 1.33, + "learning_rate": 6.73811639271435e-06, + "loss": 0.1197, "step": 11955 }, { - "epoch": 0.66, - "learning_rate": 6.735521128324727e-06, - "loss": 0.1285, + "epoch": 1.33, + "learning_rate": 6.73256330519769e-06, + "loss": 0.2622, "step": 11960 }, { - "epoch": 0.66, - "learning_rate": 6.7299683491587545e-06, - "loss": 0.1064, + "epoch": 1.33, + "learning_rate": 6.727010217681032e-06, + "loss": 0.0541, "step": 11965 }, { - "epoch": 0.66, - "learning_rate": 6.7244155699927815e-06, - "loss": 0.1304, + "epoch": 1.33, + "learning_rate": 6.721457130164372e-06, + "loss": 0.0638, "step": 11970 }, { - "epoch": 0.66, - "learning_rate": 6.718862790826809e-06, - "loss": 0.1129, + "epoch": 1.33, + "learning_rate": 6.715904042647713e-06, + "loss": 0.059, "step": 11975 }, { - "epoch": 0.66, - "learning_rate": 6.713310011660836e-06, - "loss": 0.0673, + "epoch": 1.33, + "learning_rate": 6.710350955131053e-06, + "loss": 0.1019, "step": 11980 }, { - "epoch": 0.67, - "learning_rate": 6.707757232494865e-06, - "loss": 0.0791, + "epoch": 1.33, + "learning_rate": 6.704797867614394e-06, + "loss": 0.0312, "step": 11985 }, { - "epoch": 0.67, - "learning_rate": 6.702204453328892e-06, - "loss": 0.0869, + "epoch": 1.33, + "learning_rate": 6.6992447800977344e-06, + "loss": 0.0307, "step": 11990 }, { - "epoch": 0.67, - "learning_rate": 6.69665167416292e-06, - "loss": 0.198, + "epoch": 1.33, + "learning_rate": 6.6936916925810755e-06, + "loss": 0.0525, "step": 11995 }, { - "epoch": 0.67, - "learning_rate": 6.6910988949969465e-06, - "loss": 0.3397, + "epoch": 1.33, + "learning_rate": 6.688138605064416e-06, + "loss": 0.1069, "step": 12000 }, { - "epoch": 0.67, - "eval_loss": 0.04080688953399658, - "eval_runtime": 1620.7622, - "eval_samples_per_second": 9.883, - "eval_steps_per_second": 2.471, - "eval_wer": 0.11929318427695637, + "epoch": 1.33, + "eval_loss": 0.03537467494606972, + "eval_runtime": 1821.7996, + "eval_samples_per_second": 8.792, + "eval_steps_per_second": 2.198, + "eval_wer": 0.07814641182834475, "step": 12000 }, { - "epoch": 0.67, - "learning_rate": 6.685546115830974e-06, - "loss": 0.1352, + "epoch": 1.33, + "learning_rate": 6.682585517547757e-06, + "loss": 0.1494, "step": 12005 }, { - "epoch": 0.67, - "learning_rate": 6.679993336665001e-06, - "loss": 0.0971, + "epoch": 1.33, + "learning_rate": 6.677032430031097e-06, + "loss": 0.287, "step": 12010 }, { - "epoch": 0.67, - "learning_rate": 6.674440557499029e-06, - "loss": 0.1058, + "epoch": 1.33, + "learning_rate": 6.671479342514439e-06, + "loss": 0.051, "step": 12015 }, { - "epoch": 0.67, - "learning_rate": 6.668887778333056e-06, - "loss": 0.1066, + "epoch": 1.33, + "learning_rate": 6.665926254997779e-06, + "loss": 0.0788, "step": 12020 }, { - "epoch": 0.67, - "learning_rate": 6.663334999167084e-06, - "loss": 0.0675, + "epoch": 1.33, + "learning_rate": 6.66037316748112e-06, + "loss": 0.0693, "step": 12025 }, { - "epoch": 0.67, - "learning_rate": 6.657782220001111e-06, - "loss": 0.0833, + "epoch": 1.34, + "learning_rate": 6.654820079964461e-06, + "loss": 0.0857, "step": 12030 }, { - "epoch": 0.67, - "learning_rate": 6.652229440835139e-06, - "loss": 0.0612, + "epoch": 1.34, + "learning_rate": 6.6492669924478015e-06, + "loss": 0.0516, "step": 12035 }, { - "epoch": 0.67, - "learning_rate": 6.6466766616691655e-06, - "loss": 0.1809, + "epoch": 1.34, + "learning_rate": 6.6437139049311425e-06, + "loss": 0.1087, "step": 12040 }, { - "epoch": 0.67, - "learning_rate": 6.641123882503193e-06, - "loss": 0.1474, + "epoch": 1.34, + "learning_rate": 6.638160817414483e-06, + "loss": 0.0696, "step": 12045 }, { - "epoch": 0.67, - "learning_rate": 6.63557110333722e-06, - "loss": 0.3844, + "epoch": 1.34, + "learning_rate": 6.632607729897824e-06, + "loss": 0.1212, "step": 12050 }, { - "epoch": 0.67, - "learning_rate": 6.630018324171248e-06, - "loss": 0.0831, + "epoch": 1.34, + "learning_rate": 6.627054642381164e-06, + "loss": 0.1714, "step": 12055 }, { - "epoch": 0.67, - "learning_rate": 6.624465545005275e-06, - "loss": 0.0846, + "epoch": 1.34, + "learning_rate": 6.621501554864506e-06, + "loss": 0.2119, "step": 12060 }, { - "epoch": 0.67, - "learning_rate": 6.618912765839303e-06, - "loss": 0.1064, + "epoch": 1.34, + "learning_rate": 6.615948467347846e-06, + "loss": 0.0623, "step": 12065 }, { - "epoch": 0.67, - "learning_rate": 6.61335998667333e-06, - "loss": 0.1391, + "epoch": 1.34, + "learning_rate": 6.610395379831187e-06, + "loss": 0.0845, "step": 12070 }, { - "epoch": 0.67, - "learning_rate": 6.607807207507358e-06, - "loss": 0.0847, + "epoch": 1.34, + "learning_rate": 6.6048422923145274e-06, + "loss": 0.0777, "step": 12075 }, { - "epoch": 0.67, - "learning_rate": 6.602254428341386e-06, - "loss": 0.0727, + "epoch": 1.34, + "learning_rate": 6.5992892047978685e-06, + "loss": 0.0991, "step": 12080 }, { - "epoch": 0.67, - "learning_rate": 6.596701649175413e-06, - "loss": 0.158, + "epoch": 1.34, + "learning_rate": 6.593736117281209e-06, + "loss": 0.0411, "step": 12085 }, { - "epoch": 0.67, - "learning_rate": 6.591148870009441e-06, - "loss": 0.0973, + "epoch": 1.34, + "learning_rate": 6.58818302976455e-06, + "loss": 0.0513, "step": 12090 }, - { - "epoch": 0.67, - "learning_rate": 6.585596090843468e-06, - "loss": 0.117, + { + "epoch": 1.34, + "learning_rate": 6.58262994224789e-06, + "loss": 0.0582, "step": 12095 }, { - "epoch": 0.67, - "learning_rate": 6.580043311677496e-06, - "loss": 0.3148, + "epoch": 1.34, + "learning_rate": 6.577076854731231e-06, + "loss": 0.0775, "step": 12100 }, { - "epoch": 0.67, - "learning_rate": 6.574490532511523e-06, - "loss": 0.0977, + "epoch": 1.34, + "learning_rate": 6.571523767214571e-06, + "loss": 0.1754, "step": 12105 }, { - "epoch": 0.67, - "learning_rate": 6.5689377533455504e-06, - "loss": 0.1006, + "epoch": 1.34, + "learning_rate": 6.565970679697913e-06, + "loss": 0.2479, "step": 12110 }, { - "epoch": 0.67, - "learning_rate": 6.563384974179577e-06, - "loss": 0.1157, + "epoch": 1.34, + "learning_rate": 6.5604175921812525e-06, + "loss": 0.0757, "step": 12115 }, { - "epoch": 0.67, - "learning_rate": 6.557832195013605e-06, - "loss": 0.1194, + "epoch": 1.35, + "learning_rate": 6.5548645046645944e-06, + "loss": 0.0696, "step": 12120 }, { - "epoch": 0.67, - "learning_rate": 6.552279415847632e-06, - "loss": 0.06, + "epoch": 1.35, + "learning_rate": 6.549311417147935e-06, + "loss": 0.0771, "step": 12125 }, { - "epoch": 0.67, - "learning_rate": 6.54672663668166e-06, - "loss": 0.0605, + "epoch": 1.35, + "learning_rate": 6.543758329631276e-06, + "loss": 0.0973, "step": 12130 }, { - "epoch": 0.67, - "learning_rate": 6.541173857515687e-06, - "loss": 0.119, + "epoch": 1.35, + "learning_rate": 6.538205242114616e-06, + "loss": 0.0523, "step": 12135 }, { - "epoch": 0.67, - "learning_rate": 6.535621078349715e-06, - "loss": 0.1832, + "epoch": 1.35, + "learning_rate": 6.532652154597957e-06, + "loss": 0.0466, "step": 12140 }, { - "epoch": 0.67, - "learning_rate": 6.530068299183742e-06, - "loss": 0.1399, + "epoch": 1.35, + "learning_rate": 6.527099067081297e-06, + "loss": 0.061, "step": 12145 }, { - "epoch": 0.67, - "learning_rate": 6.524515520017769e-06, - "loss": 0.3075, + "epoch": 1.35, + "learning_rate": 6.521545979564638e-06, + "loss": 0.1244, "step": 12150 }, { - "epoch": 0.67, - "learning_rate": 6.518962740851796e-06, - "loss": 0.1065, + "epoch": 1.35, + "learning_rate": 6.5159928920479785e-06, + "loss": 0.1754, "step": 12155 }, { - "epoch": 0.67, - "learning_rate": 6.513409961685824e-06, - "loss": 0.1183, + "epoch": 1.35, + "learning_rate": 6.5104398045313196e-06, + "loss": 0.2873, "step": 12160 }, { - "epoch": 0.68, - "learning_rate": 6.507857182519852e-06, - "loss": 0.1125, + "epoch": 1.35, + "learning_rate": 6.5048867170146615e-06, + "loss": 0.0712, "step": 12165 }, { - "epoch": 0.68, - "learning_rate": 6.50230440335388e-06, - "loss": 0.1564, + "epoch": 1.35, + "learning_rate": 6.499333629498002e-06, + "loss": 0.0774, "step": 12170 }, { - "epoch": 0.68, - "learning_rate": 6.496751624187907e-06, - "loss": 0.0876, + "epoch": 1.35, + "learning_rate": 6.493780541981343e-06, + "loss": 0.074, "step": 12175 }, { - "epoch": 0.68, - "learning_rate": 6.4911988450219345e-06, - "loss": 0.034, + "epoch": 1.35, + "learning_rate": 6.488227454464683e-06, + "loss": 0.0728, "step": 12180 }, { - "epoch": 0.68, - "learning_rate": 6.4856460658559614e-06, - "loss": 0.0742, + "epoch": 1.35, + "learning_rate": 6.482674366948024e-06, + "loss": 0.0781, "step": 12185 }, { - "epoch": 0.68, - "learning_rate": 6.480093286689989e-06, - "loss": 0.1472, + "epoch": 1.35, + "learning_rate": 6.477121279431364e-06, + "loss": 0.0659, "step": 12190 }, { - "epoch": 0.68, - "learning_rate": 6.474540507524016e-06, - "loss": 0.2527, + "epoch": 1.35, + "learning_rate": 6.471568191914705e-06, + "loss": 0.0685, "step": 12195 }, { - "epoch": 0.68, - "learning_rate": 6.468987728358044e-06, - "loss": 0.3212, + "epoch": 1.35, + "learning_rate": 6.4660151043980455e-06, + "loss": 0.1184, "step": 12200 }, { - "epoch": 0.68, - "learning_rate": 6.463434949192071e-06, - "loss": 0.1806, + "epoch": 1.35, + "learning_rate": 6.4604620168813866e-06, + "loss": 0.1693, "step": 12205 }, { - "epoch": 0.68, - "learning_rate": 6.457882170026099e-06, - "loss": 0.1404, + "epoch": 1.36, + "learning_rate": 6.454908929364727e-06, + "loss": 0.2137, "step": 12210 }, { - "epoch": 0.68, - "learning_rate": 6.452329390860126e-06, - "loss": 0.1209, + "epoch": 1.36, + "learning_rate": 6.449355841848069e-06, + "loss": 0.0506, "step": 12215 }, { - "epoch": 0.68, - "learning_rate": 6.4467766116941535e-06, - "loss": 0.1689, + "epoch": 1.36, + "learning_rate": 6.443802754331409e-06, + "loss": 0.0757, "step": 12220 }, { - "epoch": 0.68, - "learning_rate": 6.44122383252818e-06, - "loss": 0.0494, + "epoch": 1.36, + "learning_rate": 6.43824966681475e-06, + "loss": 0.0496, "step": 12225 }, { - "epoch": 0.68, - "learning_rate": 6.435671053362208e-06, - "loss": 0.0471, + "epoch": 1.36, + "learning_rate": 6.43269657929809e-06, + "loss": 0.0779, "step": 12230 }, { - "epoch": 0.68, - "learning_rate": 6.430118274196235e-06, - "loss": 0.1874, + "epoch": 1.36, + "learning_rate": 6.427143491781431e-06, + "loss": 0.0398, "step": 12235 }, { - "epoch": 0.68, - "learning_rate": 6.424565495030263e-06, - "loss": 0.2198, + "epoch": 1.36, + "learning_rate": 6.4215904042647715e-06, + "loss": 0.0343, "step": 12240 }, { - "epoch": 0.68, - "learning_rate": 6.41901271586429e-06, - "loss": 0.2294, + "epoch": 1.36, + "learning_rate": 6.4160373167481125e-06, + "loss": 0.0458, "step": 12245 }, { - "epoch": 0.68, - "learning_rate": 6.4134599366983185e-06, - "loss": 0.3045, + "epoch": 1.36, + "learning_rate": 6.410484229231453e-06, + "loss": 0.1117, "step": 12250 }, { - "epoch": 0.68, - "learning_rate": 6.407907157532346e-06, - "loss": 0.0807, + "epoch": 1.36, + "learning_rate": 6.404931141714794e-06, + "loss": 0.164, "step": 12255 }, { - "epoch": 0.68, - "learning_rate": 6.402354378366373e-06, - "loss": 0.0754, + "epoch": 1.36, + "learning_rate": 6.399378054198134e-06, + "loss": 0.2337, "step": 12260 }, { - "epoch": 0.68, - "learning_rate": 6.396801599200401e-06, - "loss": 0.1032, + "epoch": 1.36, + "learning_rate": 6.393824966681476e-06, + "loss": 0.0651, "step": 12265 }, { - "epoch": 0.68, - "learning_rate": 6.391248820034428e-06, - "loss": 0.135, + "epoch": 1.36, + "learning_rate": 6.388271879164816e-06, + "loss": 0.0593, "step": 12270 }, { - "epoch": 0.68, - "learning_rate": 6.385696040868455e-06, - "loss": 0.0875, + "epoch": 1.36, + "learning_rate": 6.382718791648157e-06, + "loss": 0.0968, "step": 12275 }, { - "epoch": 0.68, - "learning_rate": 6.380143261702483e-06, - "loss": 0.0504, + "epoch": 1.36, + "learning_rate": 6.3771657041314974e-06, + "loss": 0.0815, "step": 12280 }, { - "epoch": 0.68, - "learning_rate": 6.37459048253651e-06, - "loss": 0.1516, + "epoch": 1.36, + "learning_rate": 6.3716126166148385e-06, + "loss": 0.0484, "step": 12285 }, { - "epoch": 0.68, - "learning_rate": 6.3690377033705375e-06, - "loss": 0.1736, + "epoch": 1.36, + "learning_rate": 6.366059529098179e-06, + "loss": 0.0361, "step": 12290 }, { - "epoch": 0.68, - "learning_rate": 6.3634849242045645e-06, - "loss": 0.1362, + "epoch": 1.36, + "learning_rate": 6.36050644158152e-06, + "loss": 0.0624, "step": 12295 }, { - "epoch": 0.68, - "learning_rate": 6.357932145038592e-06, - "loss": 0.3305, + "epoch": 1.37, + "learning_rate": 6.354953354064861e-06, + "loss": 0.0901, "step": 12300 }, { - "epoch": 0.68, - "learning_rate": 6.352379365872619e-06, - "loss": 0.0774, + "epoch": 1.37, + "learning_rate": 6.349400266548201e-06, + "loss": 0.1173, "step": 12305 }, { - "epoch": 0.68, - "learning_rate": 6.346826586706647e-06, - "loss": 0.0943, + "epoch": 1.37, + "learning_rate": 6.343847179031543e-06, + "loss": 0.2258, "step": 12310 }, { - "epoch": 0.68, - "learning_rate": 6.341273807540674e-06, - "loss": 0.1071, + "epoch": 1.37, + "learning_rate": 6.338294091514882e-06, + "loss": 0.0756, "step": 12315 }, { - "epoch": 0.68, - "learning_rate": 6.335721028374702e-06, - "loss": 0.0998, + "epoch": 1.37, + "learning_rate": 6.332741003998224e-06, + "loss": 0.0601, "step": 12320 }, { - "epoch": 0.68, - "learning_rate": 6.330168249208729e-06, - "loss": 0.0845, + "epoch": 1.37, + "learning_rate": 6.3271879164815644e-06, + "loss": 0.0642, "step": 12325 }, { - "epoch": 0.68, - "learning_rate": 6.3246154700427565e-06, - "loss": 0.0996, + "epoch": 1.37, + "learning_rate": 6.3216348289649055e-06, + "loss": 0.1048, "step": 12330 }, { - "epoch": 0.68, - "learning_rate": 6.3190626908767835e-06, - "loss": 0.099, + "epoch": 1.37, + "learning_rate": 6.316081741448246e-06, + "loss": 0.0456, "step": 12335 }, { - "epoch": 0.68, - "learning_rate": 6.313509911710812e-06, - "loss": 0.186, + "epoch": 1.37, + "learning_rate": 6.310528653931587e-06, + "loss": 0.028, "step": 12340 }, { - "epoch": 0.69, - "learning_rate": 6.30795713254484e-06, - "loss": 0.1967, + "epoch": 1.37, + "learning_rate": 6.304975566414927e-06, + "loss": 0.0515, "step": 12345 }, { - "epoch": 0.69, - "learning_rate": 6.302404353378867e-06, - "loss": 0.5721, + "epoch": 1.37, + "learning_rate": 6.299422478898268e-06, + "loss": 0.0595, "step": 12350 }, { - "epoch": 0.69, - "learning_rate": 6.296851574212895e-06, - "loss": 0.0777, + "epoch": 1.37, + "learning_rate": 6.293869391381608e-06, + "loss": 0.1668, "step": 12355 }, { - "epoch": 0.69, - "learning_rate": 6.291298795046922e-06, - "loss": 0.1012, + "epoch": 1.37, + "learning_rate": 6.288316303864949e-06, + "loss": 0.1696, "step": 12360 }, { - "epoch": 0.69, - "learning_rate": 6.285746015880949e-06, - "loss": 0.0984, + "epoch": 1.37, + "learning_rate": 6.2827632163482896e-06, + "loss": 0.0524, "step": 12365 }, { - "epoch": 0.69, - "learning_rate": 6.280193236714976e-06, - "loss": 0.1324, + "epoch": 1.37, + "learning_rate": 6.2772101288316315e-06, + "loss": 0.0612, "step": 12370 }, { - "epoch": 0.69, - "learning_rate": 6.274640457549004e-06, - "loss": 0.0592, + "epoch": 1.37, + "learning_rate": 6.271657041314972e-06, + "loss": 0.0822, "step": 12375 }, { - "epoch": 0.69, - "learning_rate": 6.269087678383031e-06, - "loss": 0.0768, + "epoch": 1.37, + "learning_rate": 6.266103953798313e-06, + "loss": 0.0843, "step": 12380 }, { - "epoch": 0.69, - "learning_rate": 6.263534899217059e-06, - "loss": 0.0776, + "epoch": 1.37, + "learning_rate": 6.260550866281653e-06, + "loss": 0.0554, "step": 12385 }, { - "epoch": 0.69, - "learning_rate": 6.257982120051086e-06, - "loss": 0.0942, + "epoch": 1.38, + "learning_rate": 6.254997778764994e-06, + "loss": 0.0327, "step": 12390 }, { - "epoch": 0.69, - "learning_rate": 6.252429340885114e-06, - "loss": 0.1904, + "epoch": 1.38, + "learning_rate": 6.249444691248334e-06, + "loss": 0.0611, "step": 12395 }, { - "epoch": 0.69, - "learning_rate": 6.2468765617191406e-06, - "loss": 0.4873, + "epoch": 1.38, + "learning_rate": 6.243891603731675e-06, + "loss": 0.0866, "step": 12400 }, { - "epoch": 0.69, - "learning_rate": 6.241323782553168e-06, - "loss": 0.0981, + "epoch": 1.38, + "learning_rate": 6.2383385162150155e-06, + "loss": 0.1602, "step": 12405 }, { - "epoch": 0.69, - "learning_rate": 6.235771003387195e-06, - "loss": 0.1191, + "epoch": 1.38, + "learning_rate": 6.2327854286983566e-06, + "loss": 0.2472, "step": 12410 }, { - "epoch": 0.69, - "learning_rate": 6.230218224221223e-06, - "loss": 0.1249, + "epoch": 1.38, + "learning_rate": 6.227232341181697e-06, + "loss": 0.0588, "step": 12415 }, { - "epoch": 0.69, - "learning_rate": 6.22466544505525e-06, - "loss": 0.0969, + "epoch": 1.38, + "learning_rate": 6.221679253665039e-06, + "loss": 0.0697, "step": 12420 }, { - "epoch": 0.69, - "learning_rate": 6.219112665889279e-06, - "loss": 0.0689, + "epoch": 1.38, + "learning_rate": 6.216126166148379e-06, + "loss": 0.0646, "step": 12425 }, { - "epoch": 0.69, - "learning_rate": 6.213559886723306e-06, - "loss": 0.0575, + "epoch": 1.38, + "learning_rate": 6.21057307863172e-06, + "loss": 0.0877, "step": 12430 }, { - "epoch": 0.69, - "learning_rate": 6.2080071075573334e-06, - "loss": 0.1008, + "epoch": 1.38, + "learning_rate": 6.20501999111506e-06, + "loss": 0.0561, "step": 12435 }, { - "epoch": 0.69, - "learning_rate": 6.20245432839136e-06, - "loss": 0.0665, + "epoch": 1.38, + "learning_rate": 6.199466903598401e-06, + "loss": 0.0298, "step": 12440 }, { - "epoch": 0.69, - "learning_rate": 6.196901549225388e-06, - "loss": 0.2103, + "epoch": 1.38, + "learning_rate": 6.193913816081742e-06, + "loss": 0.037, "step": 12445 }, { - "epoch": 0.69, - "learning_rate": 6.191348770059415e-06, - "loss": 0.309, + "epoch": 1.38, + "learning_rate": 6.1883607285650825e-06, + "loss": 0.1196, "step": 12450 }, { - "epoch": 0.69, - "learning_rate": 6.185795990893443e-06, - "loss": 0.064, + "epoch": 1.38, + "learning_rate": 6.182807641048424e-06, + "loss": 0.1386, "step": 12455 }, { - "epoch": 0.69, - "learning_rate": 6.18024321172747e-06, - "loss": 0.1179, + "epoch": 1.38, + "learning_rate": 6.177254553531764e-06, + "loss": 0.2208, "step": 12460 }, { - "epoch": 0.69, - "learning_rate": 6.174690432561498e-06, - "loss": 0.0919, + "epoch": 1.38, + "learning_rate": 6.171701466015106e-06, + "loss": 0.0505, "step": 12465 }, { - "epoch": 0.69, - "learning_rate": 6.169137653395525e-06, - "loss": 0.103, + "epoch": 1.38, + "learning_rate": 6.166148378498445e-06, + "loss": 0.0777, "step": 12470 }, { - "epoch": 0.69, - "learning_rate": 6.163584874229552e-06, - "loss": 0.1225, + "epoch": 1.38, + "learning_rate": 6.160595290981787e-06, + "loss": 0.096, "step": 12475 }, { - "epoch": 0.69, - "learning_rate": 6.158032095063579e-06, - "loss": 0.109, + "epoch": 1.39, + "learning_rate": 6.155042203465127e-06, + "loss": 0.0811, "step": 12480 }, { - "epoch": 0.69, - "learning_rate": 6.152479315897607e-06, - "loss": 0.0863, + "epoch": 1.39, + "learning_rate": 6.149489115948468e-06, + "loss": 0.0655, "step": 12485 }, { - "epoch": 0.69, - "learning_rate": 6.146926536731634e-06, - "loss": 0.2349, + "epoch": 1.39, + "learning_rate": 6.1439360284318085e-06, + "loss": 0.0286, "step": 12490 }, { - "epoch": 0.69, - "learning_rate": 6.141373757565662e-06, - "loss": 0.1713, + "epoch": 1.39, + "learning_rate": 6.1383829409151496e-06, + "loss": 0.0557, "step": 12495 }, { - "epoch": 0.69, - "learning_rate": 6.135820978399689e-06, - "loss": 0.1853, + "epoch": 1.39, + "learning_rate": 6.13282985339849e-06, + "loss": 0.0753, "step": 12500 }, { - "epoch": 0.69, - "learning_rate": 6.130268199233717e-06, - "loss": 0.1439, + "epoch": 1.39, + "learning_rate": 6.127276765881831e-06, + "loss": 0.1109, "step": 12505 }, { - "epoch": 0.69, - "learning_rate": 6.124715420067744e-06, - "loss": 0.0826, + "epoch": 1.39, + "learning_rate": 6.121723678365171e-06, + "loss": 0.1795, "step": 12510 }, { - "epoch": 0.69, - "learning_rate": 6.119162640901772e-06, - "loss": 0.1003, + "epoch": 1.39, + "learning_rate": 6.116170590848512e-06, + "loss": 0.0494, "step": 12515 }, { - "epoch": 0.69, - "learning_rate": 6.1136098617358e-06, - "loss": 0.0835, + "epoch": 1.39, + "learning_rate": 6.110617503331852e-06, + "loss": 0.0695, "step": 12520 }, { - "epoch": 0.7, - "learning_rate": 6.108057082569827e-06, - "loss": 0.1266, + "epoch": 1.39, + "learning_rate": 6.105064415815194e-06, + "loss": 0.07, "step": 12525 }, { - "epoch": 0.7, - "learning_rate": 6.102504303403855e-06, - "loss": 0.0653, + "epoch": 1.39, + "learning_rate": 6.0995113282985344e-06, + "loss": 0.0822, "step": 12530 }, { - "epoch": 0.7, - "learning_rate": 6.096951524237882e-06, - "loss": 0.0685, + "epoch": 1.39, + "learning_rate": 6.0939582407818755e-06, + "loss": 0.0531, "step": 12535 }, { - "epoch": 0.7, - "learning_rate": 6.0913987450719095e-06, - "loss": 0.1758, + "epoch": 1.39, + "learning_rate": 6.089515770768548e-06, + "loss": 0.4155, "step": 12540 }, { - "epoch": 0.7, - "learning_rate": 6.0858459659059365e-06, - "loss": 0.1832, + "epoch": 1.39, + "learning_rate": 6.083962683251888e-06, + "loss": 0.0494, "step": 12545 }, { - "epoch": 0.7, - "learning_rate": 6.080293186739964e-06, - "loss": 0.5266, + "epoch": 1.39, + "learning_rate": 6.07840959573523e-06, + "loss": 0.1038, "step": 12550 }, { - "epoch": 0.7, - "learning_rate": 6.074740407573991e-06, - "loss": 0.1149, + "epoch": 1.39, + "learning_rate": 6.072856508218569e-06, + "loss": 0.126, "step": 12555 }, { - "epoch": 0.7, - "learning_rate": 6.069187628408019e-06, - "loss": 0.102, + "epoch": 1.39, + "learning_rate": 6.067303420701911e-06, + "loss": 0.2633, "step": 12560 }, { - "epoch": 0.7, - "learning_rate": 6.063634849242046e-06, - "loss": 0.1082, + "epoch": 1.39, + "learning_rate": 6.0617503331852515e-06, + "loss": 0.064, "step": 12565 }, { - "epoch": 0.7, - "learning_rate": 6.058082070076074e-06, - "loss": 0.1267, + "epoch": 1.4, + "learning_rate": 6.0561972456685925e-06, + "loss": 0.058, "step": 12570 }, { - "epoch": 0.7, - "learning_rate": 6.052529290910101e-06, - "loss": 0.0872, + "epoch": 1.4, + "learning_rate": 6.050644158151933e-06, + "loss": 0.0918, "step": 12575 }, { - "epoch": 0.7, - "learning_rate": 6.0469765117441285e-06, - "loss": 0.078, + "epoch": 1.4, + "learning_rate": 6.045091070635274e-06, + "loss": 0.1026, "step": 12580 }, { - "epoch": 0.7, - "learning_rate": 6.0414237325781555e-06, - "loss": 0.1548, + "epoch": 1.4, + "learning_rate": 6.039537983118614e-06, + "loss": 0.0393, "step": 12585 }, { - "epoch": 0.7, - "learning_rate": 6.035870953412183e-06, - "loss": 0.1542, + "epoch": 1.4, + "learning_rate": 6.033984895601955e-06, + "loss": 0.0523, "step": 12590 }, { - "epoch": 0.7, - "learning_rate": 6.03031817424621e-06, - "loss": 0.1944, + "epoch": 1.4, + "learning_rate": 6.028431808085295e-06, + "loss": 0.0742, "step": 12595 }, { - "epoch": 0.7, - "learning_rate": 6.024765395080238e-06, - "loss": 0.3148, + "epoch": 1.4, + "learning_rate": 6.022878720568636e-06, + "loss": 0.0861, "step": 12600 }, { - "epoch": 0.7, - "learning_rate": 6.019212615914266e-06, - "loss": 0.1503, + "epoch": 1.4, + "learning_rate": 6.017325633051977e-06, + "loss": 0.1553, "step": 12605 }, { - "epoch": 0.7, - "learning_rate": 6.013659836748294e-06, - "loss": 0.1143, + "epoch": 1.4, + "learning_rate": 6.0117725455353185e-06, + "loss": 0.279, "step": 12610 }, { - "epoch": 0.7, - "learning_rate": 6.0081070575823205e-06, - "loss": 0.0938, + "epoch": 1.4, + "learning_rate": 6.006219458018659e-06, + "loss": 0.0604, "step": 12615 }, { - "epoch": 0.7, - "learning_rate": 6.002554278416348e-06, - "loss": 0.1338, + "epoch": 1.4, + "learning_rate": 6.000666370502e-06, + "loss": 0.0825, "step": 12620 }, { - "epoch": 0.7, - "learning_rate": 5.997001499250375e-06, - "loss": 0.0474, + "epoch": 1.4, + "learning_rate": 5.99511328298534e-06, + "loss": 0.0841, "step": 12625 }, { - "epoch": 0.7, - "learning_rate": 5.991448720084403e-06, - "loss": 0.0734, + "epoch": 1.4, + "learning_rate": 5.989560195468681e-06, + "loss": 0.07, "step": 12630 }, { - "epoch": 0.7, - "learning_rate": 5.98589594091843e-06, - "loss": 0.0917, + "epoch": 1.4, + "learning_rate": 5.984007107952022e-06, + "loss": 0.0596, "step": 12635 }, { - "epoch": 0.7, - "learning_rate": 5.980343161752458e-06, - "loss": 0.1107, + "epoch": 1.4, + "learning_rate": 5.978454020435362e-06, + "loss": 0.0514, "step": 12640 }, { - "epoch": 0.7, - "learning_rate": 5.974790382586485e-06, - "loss": 0.1799, + "epoch": 1.4, + "learning_rate": 5.972900932918703e-06, + "loss": 0.0558, "step": 12645 }, { - "epoch": 0.7, - "learning_rate": 5.9692376034205126e-06, - "loss": 0.3823, + "epoch": 1.4, + "learning_rate": 5.967347845402044e-06, + "loss": 0.071, "step": 12650 }, { - "epoch": 0.7, - "learning_rate": 5.9636848242545395e-06, - "loss": 0.1439, + "epoch": 1.4, + "learning_rate": 5.9617947578853855e-06, + "loss": 0.1086, "step": 12655 }, { - "epoch": 0.7, - "learning_rate": 5.958132045088567e-06, - "loss": 0.1054, + "epoch": 1.41, + "learning_rate": 5.956241670368726e-06, + "loss": 0.2886, "step": 12660 }, { - "epoch": 0.7, - "learning_rate": 5.952579265922594e-06, - "loss": 0.0829, + "epoch": 1.41, + "learning_rate": 5.950688582852067e-06, + "loss": 0.0576, "step": 12665 }, { - "epoch": 0.7, - "learning_rate": 5.947026486756622e-06, - "loss": 0.133, + "epoch": 1.41, + "learning_rate": 5.945135495335407e-06, + "loss": 0.0563, "step": 12670 }, { - "epoch": 0.7, - "learning_rate": 5.941473707590649e-06, - "loss": 0.1351, + "epoch": 1.41, + "learning_rate": 5.939582407818748e-06, + "loss": 0.096, "step": 12675 }, { - "epoch": 0.7, - "learning_rate": 5.935920928424677e-06, - "loss": 0.0619, + "epoch": 1.41, + "learning_rate": 5.934029320302088e-06, + "loss": 0.0969, "step": 12680 }, { - "epoch": 0.7, - "learning_rate": 5.930368149258704e-06, - "loss": 0.0922, + "epoch": 1.41, + "learning_rate": 5.928476232785429e-06, + "loss": 0.0355, "step": 12685 }, { - "epoch": 0.7, - "learning_rate": 5.924815370092732e-06, - "loss": 0.1059, + "epoch": 1.41, + "learning_rate": 5.9229231452687696e-06, + "loss": 0.0617, "step": 12690 }, { - "epoch": 0.7, - "learning_rate": 5.91926259092676e-06, - "loss": 0.1501, + "epoch": 1.41, + "learning_rate": 5.917370057752111e-06, + "loss": 0.0953, "step": 12695 }, { - "epoch": 0.7, - "learning_rate": 5.913709811760787e-06, - "loss": 0.3816, + "epoch": 1.41, + "learning_rate": 5.911816970235451e-06, + "loss": 0.0747, "step": 12700 }, { - "epoch": 0.71, - "learning_rate": 5.908157032594815e-06, - "loss": 0.0588, + "epoch": 1.41, + "learning_rate": 5.906263882718793e-06, + "loss": 0.1581, "step": 12705 }, { - "epoch": 0.71, - "learning_rate": 5.902604253428842e-06, - "loss": 0.1191, + "epoch": 1.41, + "learning_rate": 5.900710795202132e-06, + "loss": 0.22, "step": 12710 }, { - "epoch": 0.71, - "learning_rate": 5.89705147426287e-06, - "loss": 0.1259, + "epoch": 1.41, + "learning_rate": 5.895157707685474e-06, + "loss": 0.054, "step": 12715 }, { - "epoch": 0.71, - "learning_rate": 5.891498695096897e-06, - "loss": 0.0916, + "epoch": 1.41, + "learning_rate": 5.889604620168814e-06, + "loss": 0.0537, "step": 12720 }, { - "epoch": 0.71, - "learning_rate": 5.885945915930924e-06, - "loss": 0.0693, + "epoch": 1.41, + "learning_rate": 5.884051532652155e-06, + "loss": 0.0727, "step": 12725 }, { - "epoch": 0.71, - "learning_rate": 5.880393136764951e-06, + "epoch": 1.41, + "learning_rate": 5.8784984451354955e-06, "loss": 0.0881, "step": 12730 }, { - "epoch": 0.71, - "learning_rate": 5.874840357598979e-06, - "loss": 0.1278, + "epoch": 1.41, + "learning_rate": 5.8729453576188366e-06, + "loss": 0.0795, "step": 12735 }, { - "epoch": 0.71, - "learning_rate": 5.869287578433006e-06, - "loss": 0.1735, + "epoch": 1.41, + "learning_rate": 5.867392270102177e-06, + "loss": 0.0766, "step": 12740 }, { - "epoch": 0.71, - "learning_rate": 5.863734799267033e-06, - "loss": 0.0867, + "epoch": 1.41, + "learning_rate": 5.861839182585518e-06, + "loss": 0.0452, "step": 12745 }, { - "epoch": 0.71, - "learning_rate": 5.858182020101061e-06, - "loss": 0.2798, + "epoch": 1.42, + "learning_rate": 5.856286095068858e-06, + "loss": 0.079, "step": 12750 }, { - "epoch": 0.71, - "learning_rate": 5.852629240935088e-06, - "loss": 0.1229, + "epoch": 1.42, + "learning_rate": 5.850733007552199e-06, + "loss": 0.1253, "step": 12755 }, { - "epoch": 0.71, - "learning_rate": 5.847076461769116e-06, - "loss": 0.0733, + "epoch": 1.42, + "learning_rate": 5.845179920035539e-06, + "loss": 0.2302, "step": 12760 }, { - "epoch": 0.71, - "learning_rate": 5.8415236826031426e-06, - "loss": 0.1169, + "epoch": 1.42, + "learning_rate": 5.839626832518881e-06, + "loss": 0.0606, "step": 12765 }, { - "epoch": 0.71, - "learning_rate": 5.83597090343717e-06, - "loss": 0.1009, + "epoch": 1.42, + "learning_rate": 5.834073745002222e-06, + "loss": 0.0561, "step": 12770 }, { - "epoch": 0.71, - "learning_rate": 5.830418124271197e-06, - "loss": 0.0505, + "epoch": 1.42, + "learning_rate": 5.8285206574855625e-06, + "loss": 0.0763, "step": 12775 }, { - "epoch": 0.71, - "learning_rate": 5.824865345105226e-06, - "loss": 0.0713, + "epoch": 1.42, + "learning_rate": 5.822967569968904e-06, + "loss": 0.0673, "step": 12780 }, { - "epoch": 0.71, - "learning_rate": 5.819312565939254e-06, - "loss": 0.1716, + "epoch": 1.42, + "learning_rate": 5.817414482452244e-06, + "loss": 0.1012, "step": 12785 }, { - "epoch": 0.71, - "learning_rate": 5.813759786773281e-06, - "loss": 0.1856, + "epoch": 1.42, + "learning_rate": 5.811861394935585e-06, + "loss": 0.0257, "step": 12790 }, { - "epoch": 0.71, - "learning_rate": 5.8082070076073085e-06, - "loss": 0.1432, + "epoch": 1.42, + "learning_rate": 5.806308307418925e-06, + "loss": 0.0518, "step": 12795 }, { - "epoch": 0.71, - "learning_rate": 5.8026542284413354e-06, - "loss": 0.3508, + "epoch": 1.42, + "learning_rate": 5.800755219902266e-06, + "loss": 0.102, "step": 12800 }, { - "epoch": 0.71, - "learning_rate": 5.797101449275363e-06, - "loss": 0.0936, + "epoch": 1.42, + "learning_rate": 5.795202132385606e-06, + "loss": 0.1302, "step": 12805 }, { - "epoch": 0.71, - "learning_rate": 5.79154867010939e-06, - "loss": 0.1398, + "epoch": 1.42, + "learning_rate": 5.789649044868948e-06, + "loss": 0.2529, "step": 12810 }, { - "epoch": 0.71, - "learning_rate": 5.785995890943418e-06, - "loss": 0.1103, + "epoch": 1.42, + "learning_rate": 5.7840959573522885e-06, + "loss": 0.0544, "step": 12815 }, { - "epoch": 0.71, - "learning_rate": 5.780443111777445e-06, - "loss": 0.1397, + "epoch": 1.42, + "learning_rate": 5.7785428698356296e-06, + "loss": 0.0657, "step": 12820 }, { - "epoch": 0.71, - "learning_rate": 5.774890332611473e-06, - "loss": 0.0849, + "epoch": 1.42, + "learning_rate": 5.77298978231897e-06, + "loss": 0.0862, "step": 12825 }, { - "epoch": 0.71, - "learning_rate": 5.7693375534455e-06, - "loss": 0.0683, + "epoch": 1.42, + "learning_rate": 5.767436694802311e-06, + "loss": 0.0707, "step": 12830 }, { - "epoch": 0.71, - "learning_rate": 5.7637847742795275e-06, - "loss": 0.0925, + "epoch": 1.42, + "learning_rate": 5.761883607285651e-06, + "loss": 0.0398, "step": 12835 }, { - "epoch": 0.71, - "learning_rate": 5.758231995113554e-06, - "loss": 0.1195, + "epoch": 1.43, + "learning_rate": 5.756330519768992e-06, + "loss": 0.0678, "step": 12840 }, { - "epoch": 0.71, - "learning_rate": 5.752679215947582e-06, - "loss": 0.188, + "epoch": 1.43, + "learning_rate": 5.750777432252332e-06, + "loss": 0.0592, "step": 12845 }, { - "epoch": 0.71, - "learning_rate": 5.747126436781609e-06, - "loss": 0.4552, + "epoch": 1.43, + "learning_rate": 5.745224344735673e-06, + "loss": 0.0633, "step": 12850 }, { - "epoch": 0.71, - "learning_rate": 5.741573657615637e-06, - "loss": 0.0776, + "epoch": 1.43, + "learning_rate": 5.739671257219014e-06, + "loss": 0.1238, "step": 12855 }, { - "epoch": 0.71, - "learning_rate": 5.736020878449664e-06, - "loss": 0.0865, + "epoch": 1.43, + "learning_rate": 5.7341181697023555e-06, + "loss": 0.2725, "step": 12860 }, { - "epoch": 0.71, - "learning_rate": 5.730468099283692e-06, - "loss": 0.0929, + "epoch": 1.43, + "learning_rate": 5.728565082185695e-06, + "loss": 0.0773, "step": 12865 }, { - "epoch": 0.71, - "learning_rate": 5.7249153201177195e-06, - "loss": 0.0914, + "epoch": 1.43, + "learning_rate": 5.723011994669037e-06, + "loss": 0.0469, "step": 12870 }, { - "epoch": 0.71, - "learning_rate": 5.719362540951747e-06, - "loss": 0.0756, + "epoch": 1.43, + "learning_rate": 5.717458907152377e-06, + "loss": 0.0613, "step": 12875 }, { - "epoch": 0.71, - "learning_rate": 5.713809761785774e-06, - "loss": 0.0648, + "epoch": 1.43, + "learning_rate": 5.711905819635718e-06, + "loss": 0.0824, "step": 12880 }, { - "epoch": 0.72, - "learning_rate": 5.708256982619802e-06, - "loss": 0.0873, + "epoch": 1.43, + "learning_rate": 5.706352732119058e-06, + "loss": 0.0737, "step": 12885 }, { - "epoch": 0.72, - "learning_rate": 5.702704203453829e-06, - "loss": 0.1125, + "epoch": 1.43, + "learning_rate": 5.700799644602399e-06, + "loss": 0.0439, "step": 12890 }, { - "epoch": 0.72, - "learning_rate": 5.697151424287857e-06, - "loss": 0.2275, + "epoch": 1.43, + "learning_rate": 5.6952465570857396e-06, + "loss": 0.061, "step": 12895 }, { - "epoch": 0.72, - "learning_rate": 5.691598645121884e-06, - "loss": 0.2478, + "epoch": 1.43, + "learning_rate": 5.689693469569081e-06, + "loss": 0.0588, "step": 12900 }, { - "epoch": 0.72, - "learning_rate": 5.6860458659559115e-06, - "loss": 0.1062, + "epoch": 1.43, + "learning_rate": 5.6841403820524225e-06, + "loss": 0.1573, "step": 12905 }, { - "epoch": 0.72, - "learning_rate": 5.6804930867899385e-06, - "loss": 0.0927, + "epoch": 1.43, + "learning_rate": 5.678587294535762e-06, + "loss": 0.2104, "step": 12910 }, { - "epoch": 0.72, - "learning_rate": 5.674940307623966e-06, - "loss": 0.0822, + "epoch": 1.43, + "learning_rate": 5.673034207019104e-06, + "loss": 0.0515, "step": 12915 }, { - "epoch": 0.72, - "learning_rate": 5.669387528457993e-06, - "loss": 0.1629, + "epoch": 1.43, + "learning_rate": 5.667481119502444e-06, + "loss": 0.0537, "step": 12920 }, { - "epoch": 0.72, - "learning_rate": 5.663834749292021e-06, - "loss": 0.0762, + "epoch": 1.43, + "learning_rate": 5.661928031985785e-06, + "loss": 0.0732, "step": 12925 }, { - "epoch": 0.72, - "learning_rate": 5.658281970126048e-06, - "loss": 0.071, + "epoch": 1.44, + "learning_rate": 5.656374944469125e-06, + "loss": 0.0755, "step": 12930 }, { - "epoch": 0.72, - "learning_rate": 5.652729190960076e-06, - "loss": 0.0631, + "epoch": 1.44, + "learning_rate": 5.650821856952466e-06, + "loss": 0.0308, "step": 12935 }, { - "epoch": 0.72, - "learning_rate": 5.647176411794103e-06, - "loss": 0.1018, + "epoch": 1.44, + "learning_rate": 5.645268769435807e-06, + "loss": 0.0568, "step": 12940 }, { - "epoch": 0.72, - "learning_rate": 5.6416236326281305e-06, - "loss": 0.1811, + "epoch": 1.44, + "learning_rate": 5.639715681919148e-06, + "loss": 0.0953, "step": 12945 }, { - "epoch": 0.72, - "learning_rate": 5.6360708534621574e-06, - "loss": 0.2982, + "epoch": 1.44, + "learning_rate": 5.634162594402488e-06, + "loss": 0.0846, "step": 12950 }, { - "epoch": 0.72, - "learning_rate": 5.630518074296186e-06, - "loss": 0.1035, + "epoch": 1.44, + "learning_rate": 5.628609506885829e-06, + "loss": 0.1323, "step": 12955 }, { - "epoch": 0.72, - "learning_rate": 5.624965295130214e-06, - "loss": 0.0923, + "epoch": 1.44, + "learning_rate": 5.623056419369169e-06, + "loss": 0.1614, "step": 12960 }, { - "epoch": 0.72, - "learning_rate": 5.619412515964241e-06, - "loss": 0.1011, + "epoch": 1.44, + "learning_rate": 5.617503331852511e-06, + "loss": 0.0802, "step": 12965 }, { - "epoch": 0.72, - "learning_rate": 5.613859736798269e-06, - "loss": 0.0904, + "epoch": 1.44, + "learning_rate": 5.611950244335851e-06, + "loss": 0.0637, "step": 12970 }, { - "epoch": 0.72, - "learning_rate": 5.6083069576322956e-06, - "loss": 0.069, + "epoch": 1.44, + "learning_rate": 5.606397156819192e-06, + "loss": 0.0508, "step": 12975 }, { - "epoch": 0.72, - "learning_rate": 5.602754178466323e-06, - "loss": 0.1125, + "epoch": 1.44, + "learning_rate": 5.6008440693025325e-06, + "loss": 0.0836, "step": 12980 }, { - "epoch": 0.72, - "learning_rate": 5.59720139930035e-06, - "loss": 0.0495, + "epoch": 1.44, + "learning_rate": 5.595290981785874e-06, + "loss": 0.0551, "step": 12985 }, { - "epoch": 0.72, - "learning_rate": 5.591648620134378e-06, - "loss": 0.1421, + "epoch": 1.44, + "learning_rate": 5.589737894269214e-06, + "loss": 0.1053, "step": 12990 }, { - "epoch": 0.72, - "learning_rate": 5.586095840968405e-06, - "loss": 0.1659, + "epoch": 1.44, + "learning_rate": 5.584184806752555e-06, + "loss": 0.0429, "step": 12995 }, { - "epoch": 0.72, - "learning_rate": 5.580543061802433e-06, - "loss": 0.3223, + "epoch": 1.44, + "learning_rate": 5.578631719235895e-06, + "loss": 0.0761, "step": 13000 }, { - "epoch": 0.72, - "learning_rate": 5.57499028263646e-06, - "loss": 0.0709, + "epoch": 1.44, + "learning_rate": 5.573078631719236e-06, + "loss": 0.1536, "step": 13005 }, { - "epoch": 0.72, - "learning_rate": 5.569437503470488e-06, - "loss": 0.0868, + "epoch": 1.44, + "learning_rate": 5.567525544202576e-06, + "loss": 0.2435, "step": 13010 }, { - "epoch": 0.72, - "learning_rate": 5.5638847243045146e-06, - "loss": 0.1106, + "epoch": 1.44, + "learning_rate": 5.561972456685918e-06, + "loss": 0.0561, "step": 13015 }, { - "epoch": 0.72, - "learning_rate": 5.558331945138542e-06, - "loss": 0.0805, + "epoch": 1.45, + "learning_rate": 5.556419369169258e-06, + "loss": 0.0656, "step": 13020 }, { - "epoch": 0.72, - "learning_rate": 5.552779165972569e-06, - "loss": 0.097, + "epoch": 1.45, + "learning_rate": 5.5508662816525996e-06, + "loss": 0.062, "step": 13025 }, { - "epoch": 0.72, - "learning_rate": 5.547226386806597e-06, - "loss": 0.0569, + "epoch": 1.45, + "learning_rate": 5.54531319413594e-06, + "loss": 0.079, "step": 13030 }, { - "epoch": 0.72, - "learning_rate": 5.541673607640624e-06, - "loss": 0.0779, + "epoch": 1.45, + "learning_rate": 5.539760106619281e-06, + "loss": 0.067, "step": 13035 }, { - "epoch": 0.72, - "learning_rate": 5.536120828474652e-06, - "loss": 0.1514, + "epoch": 1.45, + "learning_rate": 5.534207019102622e-06, + "loss": 0.0296, "step": 13040 }, { - "epoch": 0.72, - "learning_rate": 5.53056804930868e-06, - "loss": 0.1261, + "epoch": 1.45, + "learning_rate": 5.528653931585962e-06, + "loss": 0.1216, "step": 13045 }, { - "epoch": 0.72, - "learning_rate": 5.5250152701427074e-06, - "loss": 0.2832, + "epoch": 1.45, + "learning_rate": 5.523100844069303e-06, + "loss": 0.0791, "step": 13050 }, { - "epoch": 0.72, - "learning_rate": 5.519462490976734e-06, - "loss": 0.0668, + "epoch": 1.45, + "learning_rate": 5.517547756552643e-06, + "loss": 0.1093, "step": 13055 }, { - "epoch": 0.72, - "learning_rate": 5.513909711810762e-06, - "loss": 0.0899, + "epoch": 1.45, + "learning_rate": 5.511994669035985e-06, + "loss": 0.2942, "step": 13060 }, { - "epoch": 0.73, - "learning_rate": 5.508356932644789e-06, - "loss": 0.1231, + "epoch": 1.45, + "learning_rate": 5.506441581519325e-06, + "loss": 0.0634, "step": 13065 }, { - "epoch": 0.73, - "learning_rate": 5.502804153478817e-06, - "loss": 0.1447, + "epoch": 1.45, + "learning_rate": 5.5008884940026666e-06, + "loss": 0.0771, "step": 13070 }, { - "epoch": 0.73, - "learning_rate": 5.497251374312844e-06, - "loss": 0.08, + "epoch": 1.45, + "learning_rate": 5.495335406486007e-06, + "loss": 0.0598, "step": 13075 }, { - "epoch": 0.73, - "learning_rate": 5.491698595146872e-06, - "loss": 0.0423, + "epoch": 1.45, + "learning_rate": 5.489782318969348e-06, + "loss": 0.0798, "step": 13080 }, { - "epoch": 0.73, - "learning_rate": 5.486145815980899e-06, - "loss": 0.0635, + "epoch": 1.45, + "learning_rate": 5.484229231452688e-06, + "loss": 0.0405, "step": 13085 }, { - "epoch": 0.73, - "learning_rate": 5.480593036814926e-06, - "loss": 0.1268, + "epoch": 1.45, + "learning_rate": 5.478676143936029e-06, + "loss": 0.0699, "step": 13090 }, { - "epoch": 0.73, - "learning_rate": 5.475040257648953e-06, - "loss": 0.1337, + "epoch": 1.45, + "learning_rate": 5.473123056419369e-06, + "loss": 0.0763, "step": 13095 }, { - "epoch": 0.73, - "learning_rate": 5.469487478482981e-06, - "loss": 0.2652, + "epoch": 1.45, + "learning_rate": 5.46756996890271e-06, + "loss": 0.0934, "step": 13100 }, { - "epoch": 0.73, - "learning_rate": 5.463934699317008e-06, - "loss": 0.1026, + "epoch": 1.45, + "learning_rate": 5.462016881386051e-06, + "loss": 0.1859, "step": 13105 }, { - "epoch": 0.73, - "learning_rate": 5.458381920151036e-06, - "loss": 0.0925, + "epoch": 1.46, + "learning_rate": 5.456463793869392e-06, + "loss": 0.2356, "step": 13110 }, { - "epoch": 0.73, - "learning_rate": 5.452829140985063e-06, - "loss": 0.0927, + "epoch": 1.46, + "learning_rate": 5.450910706352732e-06, + "loss": 0.0738, "step": 13115 }, { - "epoch": 0.73, - "learning_rate": 5.447276361819091e-06, - "loss": 0.1275, + "epoch": 1.46, + "learning_rate": 5.445357618836074e-06, + "loss": 0.0646, "step": 13120 }, { - "epoch": 0.73, - "learning_rate": 5.441723582653118e-06, - "loss": 0.1294, + "epoch": 1.46, + "learning_rate": 5.439804531319414e-06, + "loss": 0.0638, "step": 13125 }, { - "epoch": 0.73, - "learning_rate": 5.436170803487145e-06, - "loss": 0.0438, + "epoch": 1.46, + "learning_rate": 5.434251443802755e-06, + "loss": 0.0923, "step": 13130 }, { - "epoch": 0.73, - "learning_rate": 5.430618024321174e-06, - "loss": 0.0767, + "epoch": 1.46, + "learning_rate": 5.428698356286095e-06, + "loss": 0.0363, "step": 13135 }, { - "epoch": 0.73, - "learning_rate": 5.425065245155201e-06, - "loss": 0.0874, + "epoch": 1.46, + "learning_rate": 5.423145268769436e-06, + "loss": 0.0294, "step": 13140 }, { - "epoch": 0.73, - "learning_rate": 5.419512465989229e-06, - "loss": 0.191, + "epoch": 1.46, + "learning_rate": 5.417592181252777e-06, + "loss": 0.0375, "step": 13145 }, { - "epoch": 0.73, - "learning_rate": 5.413959686823256e-06, - "loss": 0.2695, + "epoch": 1.46, + "learning_rate": 5.412039093736118e-06, + "loss": 0.0951, "step": 13150 }, { - "epoch": 0.73, - "learning_rate": 5.4084069076572835e-06, - "loss": 0.0806, + "epoch": 1.46, + "learning_rate": 5.406486006219458e-06, + "loss": 0.1879, "step": 13155 }, { - "epoch": 0.73, - "learning_rate": 5.4028541284913105e-06, - "loss": 0.097, + "epoch": 1.46, + "learning_rate": 5.400932918702799e-06, + "loss": 0.1814, "step": 13160 }, { - "epoch": 0.73, - "learning_rate": 5.397301349325338e-06, - "loss": 0.12, + "epoch": 1.46, + "learning_rate": 5.395379831186139e-06, + "loss": 0.0906, "step": 13165 }, { - "epoch": 0.73, - "learning_rate": 5.391748570159365e-06, - "loss": 0.1272, + "epoch": 1.46, + "learning_rate": 5.389826743669481e-06, + "loss": 0.0575, "step": 13170 }, { - "epoch": 0.73, - "learning_rate": 5.386195790993393e-06, - "loss": 0.0842, + "epoch": 1.46, + "learning_rate": 5.384273656152822e-06, + "loss": 0.0752, "step": 13175 }, { - "epoch": 0.73, - "learning_rate": 5.38064301182742e-06, - "loss": 0.0431, + "epoch": 1.46, + "learning_rate": 5.378720568636162e-06, + "loss": 0.0898, "step": 13180 }, { - "epoch": 0.73, - "learning_rate": 5.375090232661448e-06, - "loss": 0.0696, + "epoch": 1.46, + "learning_rate": 5.373167481119503e-06, + "loss": 0.0563, "step": 13185 }, { - "epoch": 0.73, - "learning_rate": 5.369537453495475e-06, - "loss": 0.1279, + "epoch": 1.46, + "learning_rate": 5.367614393602844e-06, + "loss": 0.0496, "step": 13190 }, { - "epoch": 0.73, - "learning_rate": 5.3639846743295025e-06, - "loss": 0.1745, + "epoch": 1.46, + "learning_rate": 5.362061306086185e-06, + "loss": 0.0737, "step": 13195 }, { - "epoch": 0.73, - "learning_rate": 5.3584318951635294e-06, - "loss": 0.3915, + "epoch": 1.47, + "learning_rate": 5.356508218569525e-06, + "loss": 0.1588, "step": 13200 }, { - "epoch": 0.73, - "learning_rate": 5.352879115997557e-06, - "loss": 0.0834, + "epoch": 1.47, + "learning_rate": 5.350955131052866e-06, + "loss": 0.1223, "step": 13205 }, { - "epoch": 0.73, - "learning_rate": 5.347326336831584e-06, - "loss": 0.0939, + "epoch": 1.47, + "learning_rate": 5.345402043536206e-06, + "loss": 0.2363, "step": 13210 }, { - "epoch": 0.73, - "learning_rate": 5.341773557665612e-06, - "loss": 0.0818, + "epoch": 1.47, + "learning_rate": 5.339848956019548e-06, + "loss": 0.0529, "step": 13215 }, { - "epoch": 0.73, - "learning_rate": 5.33622077849964e-06, - "loss": 0.1018, + "epoch": 1.47, + "learning_rate": 5.3342958685028874e-06, + "loss": 0.0543, "step": 13220 }, { - "epoch": 0.73, - "learning_rate": 5.3306679993336676e-06, - "loss": 0.0904, + "epoch": 1.47, + "learning_rate": 5.328742780986229e-06, + "loss": 0.0746, "step": 13225 }, { - "epoch": 0.73, - "learning_rate": 5.3251152201676945e-06, - "loss": 0.0682, + "epoch": 1.47, + "learning_rate": 5.3231896934695696e-06, + "loss": 0.0751, "step": 13230 }, { - "epoch": 0.73, - "learning_rate": 5.319562441001722e-06, - "loss": 0.072, + "epoch": 1.47, + "learning_rate": 5.317636605952911e-06, + "loss": 0.0679, "step": 13235 }, { - "epoch": 0.73, - "learning_rate": 5.314009661835749e-06, - "loss": 0.1775, + "epoch": 1.47, + "learning_rate": 5.312083518436251e-06, + "loss": 0.0339, "step": 13240 }, { - "epoch": 0.74, - "learning_rate": 5.308456882669777e-06, - "loss": 0.1895, + "epoch": 1.47, + "learning_rate": 5.306530430919592e-06, + "loss": 0.1017, "step": 13245 }, { - "epoch": 0.74, - "learning_rate": 5.302904103503804e-06, - "loss": 0.2996, + "epoch": 1.47, + "learning_rate": 5.300977343402932e-06, + "loss": 0.0852, "step": 13250 }, { - "epoch": 0.74, - "learning_rate": 5.297351324337832e-06, - "loss": 0.0734, + "epoch": 1.47, + "learning_rate": 5.295424255886273e-06, + "loss": 0.1288, "step": 13255 }, { - "epoch": 0.74, - "learning_rate": 5.291798545171859e-06, - "loss": 0.0934, + "epoch": 1.47, + "learning_rate": 5.289871168369613e-06, + "loss": 0.2528, "step": 13260 }, { - "epoch": 0.74, - "learning_rate": 5.2862457660058866e-06, - "loss": 0.1039, + "epoch": 1.47, + "learning_rate": 5.2843180808529545e-06, + "loss": 0.0578, "step": 13265 }, { - "epoch": 0.74, - "learning_rate": 5.2806929868399135e-06, - "loss": 0.1092, + "epoch": 1.47, + "learning_rate": 5.278764993336295e-06, + "loss": 0.0495, "step": 13270 }, { - "epoch": 0.74, - "learning_rate": 5.275140207673941e-06, - "loss": 0.1258, + "epoch": 1.47, + "learning_rate": 5.2732119058196366e-06, + "loss": 0.0798, "step": 13275 }, { - "epoch": 0.74, - "learning_rate": 5.269587428507968e-06, - "loss": 0.1327, + "epoch": 1.47, + "learning_rate": 5.267658818302977e-06, + "loss": 0.0888, "step": 13280 }, { - "epoch": 0.74, - "learning_rate": 5.264034649341996e-06, - "loss": 0.1103, + "epoch": 1.47, + "learning_rate": 5.262105730786318e-06, + "loss": 0.0335, "step": 13285 }, { - "epoch": 0.74, - "learning_rate": 5.258481870176023e-06, - "loss": 0.1438, + "epoch": 1.48, + "learning_rate": 5.256552643269658e-06, + "loss": 0.0405, "step": 13290 }, { - "epoch": 0.74, - "learning_rate": 5.254039646843245e-06, - "loss": 0.5745, + "epoch": 1.48, + "learning_rate": 5.250999555752999e-06, + "loss": 0.073, "step": 13295 }, { - "epoch": 0.74, - "learning_rate": 5.248486867677273e-06, - "loss": 0.2957, + "epoch": 1.48, + "learning_rate": 5.245446468236339e-06, + "loss": 0.1129, "step": 13300 }, { - "epoch": 0.74, - "learning_rate": 5.2429340885113e-06, - "loss": 0.1027, + "epoch": 1.48, + "learning_rate": 5.23989338071968e-06, + "loss": 0.1393, "step": 13305 }, { - "epoch": 0.74, - "learning_rate": 5.237381309345328e-06, - "loss": 0.0896, + "epoch": 1.48, + "learning_rate": 5.2343402932030215e-06, + "loss": 0.2412, "step": 13310 }, { - "epoch": 0.74, - "learning_rate": 5.2318285301793545e-06, - "loss": 0.1206, + "epoch": 1.48, + "learning_rate": 5.228787205686362e-06, + "loss": 0.0574, "step": 13315 }, { - "epoch": 0.74, - "learning_rate": 5.226275751013382e-06, - "loss": 0.104, + "epoch": 1.48, + "learning_rate": 5.223234118169704e-06, + "loss": 0.0687, "step": 13320 }, { - "epoch": 0.74, - "learning_rate": 5.220722971847409e-06, - "loss": 0.075, + "epoch": 1.48, + "learning_rate": 5.217681030653044e-06, + "loss": 0.0642, "step": 13325 }, { - "epoch": 0.74, - "learning_rate": 5.215170192681438e-06, - "loss": 0.0182, + "epoch": 1.48, + "learning_rate": 5.212127943136385e-06, + "loss": 0.0731, "step": 13330 }, { - "epoch": 0.74, - "learning_rate": 5.209617413515466e-06, - "loss": 0.1027, + "epoch": 1.48, + "learning_rate": 5.206574855619725e-06, + "loss": 0.0379, "step": 13335 }, { - "epoch": 0.74, - "learning_rate": 5.204064634349493e-06, - "loss": 0.0957, + "epoch": 1.48, + "learning_rate": 5.201021768103066e-06, + "loss": 0.0405, "step": 13340 }, { - "epoch": 0.74, - "learning_rate": 5.1985118551835205e-06, - "loss": 0.226, + "epoch": 1.48, + "learning_rate": 5.195468680586406e-06, + "loss": 0.0416, "step": 13345 }, { - "epoch": 0.74, - "learning_rate": 5.192959076017547e-06, - "loss": 0.2457, + "epoch": 1.48, + "learning_rate": 5.1899155930697474e-06, + "loss": 0.0693, "step": 13350 }, { - "epoch": 0.74, - "learning_rate": 5.187406296851575e-06, - "loss": 0.0941, + "epoch": 1.48, + "learning_rate": 5.184362505553088e-06, + "loss": 0.1614, "step": 13355 }, { - "epoch": 0.74, - "learning_rate": 5.181853517685602e-06, - "loss": 0.1014, + "epoch": 1.48, + "learning_rate": 5.178809418036429e-06, + "loss": 0.2359, "step": 13360 }, { - "epoch": 0.74, - "learning_rate": 5.17630073851963e-06, - "loss": 0.0965, + "epoch": 1.48, + "learning_rate": 5.173256330519769e-06, + "loss": 0.049, "step": 13365 }, { - "epoch": 0.74, - "learning_rate": 5.170747959353657e-06, - "loss": 0.1003, + "epoch": 1.48, + "learning_rate": 5.167703243003111e-06, + "loss": 0.0714, "step": 13370 }, { - "epoch": 0.74, - "learning_rate": 5.165195180187685e-06, - "loss": 0.1301, + "epoch": 1.48, + "learning_rate": 5.16215015548645e-06, + "loss": 0.0754, "step": 13375 }, { - "epoch": 0.74, - "learning_rate": 5.159642401021712e-06, - "loss": 0.0182, + "epoch": 1.49, + "learning_rate": 5.156597067969792e-06, + "loss": 0.1066, "step": 13380 }, { - "epoch": 0.74, - "learning_rate": 5.154089621855739e-06, - "loss": 0.1064, + "epoch": 1.49, + "learning_rate": 5.151043980453132e-06, + "loss": 0.0325, "step": 13385 }, { - "epoch": 0.74, - "learning_rate": 5.148536842689766e-06, - "loss": 0.1026, + "epoch": 1.49, + "learning_rate": 5.145490892936473e-06, + "loss": 0.0861, "step": 13390 }, { - "epoch": 0.74, - "learning_rate": 5.142984063523793e-06, - "loss": 0.1922, + "epoch": 1.49, + "learning_rate": 5.139937805419814e-06, + "loss": 0.0466, "step": 13395 }, { - "epoch": 0.74, - "learning_rate": 5.137431284357821e-06, - "loss": 0.4219, + "epoch": 1.49, + "learning_rate": 5.134384717903155e-06, + "loss": 0.132, "step": 13400 }, { - "epoch": 0.74, - "learning_rate": 5.131878505191848e-06, - "loss": 0.0979, + "epoch": 1.49, + "learning_rate": 5.128831630386495e-06, + "loss": 0.1693, "step": 13405 }, { - "epoch": 0.74, - "learning_rate": 5.126325726025876e-06, - "loss": 0.0748, + "epoch": 1.49, + "learning_rate": 5.123278542869836e-06, + "loss": 0.2495, "step": 13410 }, { - "epoch": 0.74, - "learning_rate": 5.1207729468599045e-06, - "loss": 0.1213, + "epoch": 1.49, + "learning_rate": 5.117725455353176e-06, + "loss": 0.0579, "step": 13415 }, { - "epoch": 0.74, - "learning_rate": 5.1152201676939315e-06, - "loss": 0.1147, + "epoch": 1.49, + "learning_rate": 5.112172367836517e-06, + "loss": 0.0703, "step": 13420 }, { - "epoch": 0.75, - "learning_rate": 5.109667388527959e-06, - "loss": 0.1159, + "epoch": 1.49, + "learning_rate": 5.1066192803198574e-06, + "loss": 0.1037, "step": 13425 }, { - "epoch": 0.75, - "learning_rate": 5.104114609361986e-06, - "loss": 0.05, + "epoch": 1.49, + "learning_rate": 5.101066192803199e-06, + "loss": 0.0837, "step": 13430 }, { - "epoch": 0.75, - "learning_rate": 5.098561830196014e-06, - "loss": 0.1953, + "epoch": 1.49, + "learning_rate": 5.0955131052865396e-06, + "loss": 0.0601, "step": 13435 }, { - "epoch": 0.75, - "learning_rate": 5.093009051030041e-06, - "loss": 0.058, + "epoch": 1.49, + "learning_rate": 5.089960017769881e-06, + "loss": 0.0532, "step": 13440 }, { - "epoch": 0.75, - "learning_rate": 5.087456271864069e-06, - "loss": 0.1263, + "epoch": 1.49, + "learning_rate": 5.084406930253222e-06, + "loss": 0.0667, "step": 13445 }, { - "epoch": 0.75, - "learning_rate": 5.081903492698096e-06, - "loss": 0.2974, + "epoch": 1.49, + "learning_rate": 5.078853842736562e-06, + "loss": 0.0986, "step": 13450 }, { - "epoch": 0.75, - "learning_rate": 5.0763507135321235e-06, - "loss": 0.0731, + "epoch": 1.49, + "learning_rate": 5.073300755219903e-06, + "loss": 0.1116, "step": 13455 }, { - "epoch": 0.75, - "learning_rate": 5.0707979343661505e-06, - "loss": 0.1293, + "epoch": 1.49, + "learning_rate": 5.067747667703243e-06, + "loss": 0.2368, "step": 13460 }, { - "epoch": 0.75, - "learning_rate": 5.065245155200178e-06, - "loss": 0.0668, + "epoch": 1.49, + "learning_rate": 5.062194580186584e-06, + "loss": 0.0586, "step": 13465 }, { - "epoch": 0.75, - "learning_rate": 5.059692376034205e-06, - "loss": 0.0807, + "epoch": 1.5, + "learning_rate": 5.0566414926699245e-06, + "loss": 0.0654, "step": 13470 }, { - "epoch": 0.75, - "learning_rate": 5.054139596868233e-06, - "loss": 0.1141, + "epoch": 1.5, + "learning_rate": 5.051088405153266e-06, + "loss": 0.0871, "step": 13475 }, { - "epoch": 0.75, - "learning_rate": 5.04858681770226e-06, - "loss": 0.0448, + "epoch": 1.5, + "learning_rate": 5.045535317636607e-06, + "loss": 0.0598, "step": 13480 }, { - "epoch": 0.75, - "learning_rate": 5.043034038536288e-06, - "loss": 0.0757, + "epoch": 1.5, + "learning_rate": 5.039982230119948e-06, + "loss": 0.065, "step": 13485 }, { - "epoch": 0.75, - "learning_rate": 5.037481259370315e-06, - "loss": 0.1005, + "epoch": 1.5, + "learning_rate": 5.034429142603288e-06, + "loss": 0.0366, "step": 13490 }, { - "epoch": 0.75, - "learning_rate": 5.0319284802043425e-06, - "loss": 0.2119, + "epoch": 1.5, + "learning_rate": 5.028876055086629e-06, + "loss": 0.08, "step": 13495 }, { - "epoch": 0.75, - "learning_rate": 5.0263757010383694e-06, - "loss": 0.3061, + "epoch": 1.5, + "learning_rate": 5.023322967569969e-06, + "loss": 0.0897, "step": 13500 }, { - "epoch": 0.75, - "learning_rate": 5.020822921872398e-06, - "loss": 0.0779, + "epoch": 1.5, + "learning_rate": 5.01776988005331e-06, + "loss": 0.1548, "step": 13505 }, { - "epoch": 0.75, - "learning_rate": 5.015270142706425e-06, - "loss": 0.098, + "epoch": 1.5, + "learning_rate": 5.01221679253665e-06, + "loss": 0.3389, "step": 13510 }, { - "epoch": 0.75, - "learning_rate": 5.009717363540453e-06, - "loss": 0.1658, + "epoch": 1.5, + "learning_rate": 5.0066637050199915e-06, + "loss": 0.0602, "step": 13515 }, { - "epoch": 0.75, - "learning_rate": 5.00416458437448e-06, - "loss": 0.0874, + "epoch": 1.5, + "learning_rate": 5.001110617503332e-06, + "loss": 0.0744, "step": 13520 }, { - "epoch": 0.75, - "learning_rate": 4.9986118052085076e-06, - "loss": 0.0535, + "epoch": 1.5, + "learning_rate": 4.995557529986674e-06, + "loss": 0.0692, "step": 13525 }, { - "epoch": 0.75, - "learning_rate": 4.9930590260425345e-06, - "loss": 0.0886, + "epoch": 1.5, + "learning_rate": 4.990004442470014e-06, + "loss": 0.0759, "step": 13530 }, { - "epoch": 0.75, - "learning_rate": 4.987506246876562e-06, - "loss": 0.1051, + "epoch": 1.5, + "learning_rate": 4.984451354953355e-06, + "loss": 0.0509, "step": 13535 }, { - "epoch": 0.75, - "learning_rate": 4.981953467710589e-06, - "loss": 0.0757, + "epoch": 1.5, + "learning_rate": 4.978898267436695e-06, + "loss": 0.0206, "step": 13540 }, { - "epoch": 0.75, - "learning_rate": 4.976400688544617e-06, - "loss": 0.2466, + "epoch": 1.5, + "learning_rate": 4.973345179920036e-06, + "loss": 0.0467, "step": 13545 }, { - "epoch": 0.75, - "learning_rate": 4.970847909378644e-06, - "loss": 0.4621, + "epoch": 1.5, + "learning_rate": 4.967792092403376e-06, + "loss": 0.0769, "step": 13550 }, { - "epoch": 0.75, - "learning_rate": 4.965295130212672e-06, - "loss": 0.0932, + "epoch": 1.5, + "learning_rate": 4.9622390048867174e-06, + "loss": 0.1698, "step": 13555 }, { - "epoch": 0.75, - "learning_rate": 4.959742351046699e-06, - "loss": 0.1006, + "epoch": 1.51, + "learning_rate": 4.9566859173700585e-06, + "loss": 0.227, "step": 13560 }, { - "epoch": 0.75, - "learning_rate": 4.9541895718807265e-06, - "loss": 0.0856, + "epoch": 1.51, + "learning_rate": 4.951132829853399e-06, + "loss": 0.0708, "step": 13565 }, { - "epoch": 0.75, - "learning_rate": 4.948636792714754e-06, - "loss": 0.1303, + "epoch": 1.51, + "learning_rate": 4.94557974233674e-06, + "loss": 0.0702, "step": 13570 }, { - "epoch": 0.75, - "learning_rate": 4.943084013548782e-06, - "loss": 0.0661, + "epoch": 1.51, + "learning_rate": 4.94002665482008e-06, + "loss": 0.0766, "step": 13575 }, { - "epoch": 0.75, - "learning_rate": 4.937531234382809e-06, - "loss": 0.0471, + "epoch": 1.51, + "learning_rate": 4.934473567303421e-06, + "loss": 0.0652, "step": 13580 }, { - "epoch": 0.75, - "learning_rate": 4.931978455216837e-06, - "loss": 0.1427, + "epoch": 1.51, + "learning_rate": 4.928920479786762e-06, + "loss": 0.0571, "step": 13585 }, { - "epoch": 0.75, - "learning_rate": 4.926425676050864e-06, - "loss": 0.0835, + "epoch": 1.51, + "learning_rate": 4.923367392270102e-06, + "loss": 0.0648, "step": 13590 }, { - "epoch": 0.75, - "learning_rate": 4.920872896884892e-06, - "loss": 0.258, + "epoch": 1.51, + "learning_rate": 4.917814304753443e-06, + "loss": 0.0454, "step": 13595 }, { - "epoch": 0.75, - "learning_rate": 4.915320117718919e-06, - "loss": 0.413, + "epoch": 1.51, + "learning_rate": 4.912261217236784e-06, + "loss": 0.1342, "step": 13600 }, { - "epoch": 0.76, - "learning_rate": 4.909767338552946e-06, - "loss": 0.0962, + "epoch": 1.51, + "learning_rate": 4.906708129720125e-06, + "loss": 0.1654, "step": 13605 }, { - "epoch": 0.76, - "learning_rate": 4.904214559386973e-06, - "loss": 0.0812, + "epoch": 1.51, + "learning_rate": 4.901155042203466e-06, + "loss": 0.2228, "step": 13610 }, { - "epoch": 0.76, - "learning_rate": 4.898661780221001e-06, - "loss": 0.1145, + "epoch": 1.51, + "learning_rate": 4.895601954686806e-06, + "loss": 0.0578, "step": 13615 }, { - "epoch": 0.76, - "learning_rate": 4.893109001055029e-06, - "loss": 0.1613, + "epoch": 1.51, + "learning_rate": 4.890048867170147e-06, + "loss": 0.0548, "step": 13620 }, { - "epoch": 0.76, - "learning_rate": 4.887556221889056e-06, - "loss": 0.0947, + "epoch": 1.51, + "learning_rate": 4.884495779653487e-06, + "loss": 0.0941, "step": 13625 }, { - "epoch": 0.76, - "learning_rate": 4.882003442723084e-06, - "loss": 0.129, + "epoch": 1.51, + "learning_rate": 4.878942692136828e-06, + "loss": 0.0875, "step": 13630 }, { - "epoch": 0.76, - "learning_rate": 4.876450663557111e-06, - "loss": 0.1132, + "epoch": 1.51, + "learning_rate": 4.873389604620169e-06, + "loss": 0.0203, "step": 13635 }, { - "epoch": 0.76, - "learning_rate": 4.870897884391138e-06, - "loss": 0.0981, + "epoch": 1.51, + "learning_rate": 4.8678365171035096e-06, + "loss": 0.0259, "step": 13640 }, { - "epoch": 0.76, - "learning_rate": 4.865345105225165e-06, - "loss": 0.1263, + "epoch": 1.51, + "learning_rate": 4.862283429586851e-06, + "loss": 0.0657, "step": 13645 }, { - "epoch": 0.76, - "learning_rate": 4.859792326059193e-06, - "loss": 0.3383, + "epoch": 1.52, + "learning_rate": 4.856730342070192e-06, + "loss": 0.0903, "step": 13650 }, { - "epoch": 0.76, - "learning_rate": 4.85423954689322e-06, - "loss": 0.1254, + "epoch": 1.52, + "learning_rate": 4.851177254553533e-06, + "loss": 0.0966, "step": 13655 }, { - "epoch": 0.76, - "learning_rate": 4.848686767727248e-06, - "loss": 0.1254, + "epoch": 1.52, + "learning_rate": 4.845624167036873e-06, + "loss": 0.2744, "step": 13660 }, { - "epoch": 0.76, - "learning_rate": 4.843133988561276e-06, - "loss": 0.1199, + "epoch": 1.52, + "learning_rate": 4.840071079520214e-06, + "loss": 0.0586, "step": 13665 }, { - "epoch": 0.76, - "learning_rate": 4.837581209395303e-06, - "loss": 0.117, + "epoch": 1.52, + "learning_rate": 4.834517992003554e-06, + "loss": 0.0759, "step": 13670 }, { - "epoch": 0.76, - "learning_rate": 4.8320284302293304e-06, - "loss": 0.0656, + "epoch": 1.52, + "learning_rate": 4.828964904486895e-06, + "loss": 0.0834, "step": 13675 }, { - "epoch": 0.76, - "learning_rate": 4.826475651063357e-06, - "loss": 0.0425, + "epoch": 1.52, + "learning_rate": 4.823411816970236e-06, + "loss": 0.0909, "step": 13680 }, { - "epoch": 0.76, - "learning_rate": 4.820922871897385e-06, - "loss": 0.0995, + "epoch": 1.52, + "learning_rate": 4.817858729453577e-06, + "loss": 0.0378, "step": 13685 }, { - "epoch": 0.76, - "learning_rate": 4.815370092731412e-06, - "loss": 0.1156, + "epoch": 1.52, + "learning_rate": 4.812305641936918e-06, + "loss": 0.0379, "step": 13690 }, { - "epoch": 0.76, - "learning_rate": 4.80981731356544e-06, - "loss": 0.2334, + "epoch": 1.52, + "learning_rate": 4.806752554420258e-06, + "loss": 0.0712, "step": 13695 }, { - "epoch": 0.76, - "learning_rate": 4.804264534399467e-06, - "loss": 0.136, + "epoch": 1.52, + "learning_rate": 4.801199466903599e-06, + "loss": 0.1161, "step": 13700 }, { - "epoch": 0.76, - "learning_rate": 4.798711755233495e-06, - "loss": 0.0749, + "epoch": 1.52, + "learning_rate": 4.795646379386939e-06, + "loss": 0.1397, "step": 13705 }, { - "epoch": 0.76, - "learning_rate": 4.7931589760675225e-06, - "loss": 0.0836, + "epoch": 1.52, + "learning_rate": 4.79009329187028e-06, + "loss": 0.3017, "step": 13710 }, { - "epoch": 0.76, - "learning_rate": 4.787606196901549e-06, - "loss": 0.0903, + "epoch": 1.52, + "learning_rate": 4.784540204353621e-06, + "loss": 0.0687, "step": 13715 }, { - "epoch": 0.76, - "learning_rate": 4.782053417735577e-06, - "loss": 0.0892, + "epoch": 1.52, + "learning_rate": 4.7789871168369615e-06, + "loss": 0.0664, "step": 13720 }, { - "epoch": 0.76, - "learning_rate": 4.776500638569604e-06, - "loss": 0.1064, + "epoch": 1.52, + "learning_rate": 4.7734340293203025e-06, + "loss": 0.0766, "step": 13725 }, { - "epoch": 0.76, - "learning_rate": 4.770947859403632e-06, - "loss": 0.0598, + "epoch": 1.52, + "learning_rate": 4.767880941803643e-06, + "loss": 0.0785, "step": 13730 }, { - "epoch": 0.76, - "learning_rate": 4.765395080237659e-06, - "loss": 0.1061, + "epoch": 1.52, + "learning_rate": 4.762327854286984e-06, + "loss": 0.0572, "step": 13735 }, { - "epoch": 0.76, - "learning_rate": 4.759842301071687e-06, - "loss": 0.1264, + "epoch": 1.53, + "learning_rate": 4.756774766770325e-06, + "loss": 0.0619, "step": 13740 }, { - "epoch": 0.76, - "learning_rate": 4.7542895219057145e-06, - "loss": 0.2359, + "epoch": 1.53, + "learning_rate": 4.751221679253665e-06, + "loss": 0.0543, "step": 13745 }, { - "epoch": 0.76, - "learning_rate": 4.7487367427397414e-06, - "loss": 0.2118, + "epoch": 1.53, + "learning_rate": 4.745668591737006e-06, + "loss": 0.1095, "step": 13750 }, { - "epoch": 0.76, - "learning_rate": 4.743183963573769e-06, - "loss": 0.1069, + "epoch": 1.53, + "learning_rate": 4.740115504220346e-06, + "loss": 0.1272, "step": 13755 }, { - "epoch": 0.76, - "learning_rate": 4.737631184407796e-06, - "loss": 0.1058, + "epoch": 1.53, + "learning_rate": 4.7345624167036874e-06, + "loss": 0.2245, "step": 13760 }, { - "epoch": 0.76, - "learning_rate": 4.732078405241824e-06, - "loss": 0.1107, + "epoch": 1.53, + "learning_rate": 4.7290093291870285e-06, + "loss": 0.0494, "step": 13765 }, { - "epoch": 0.76, - "learning_rate": 4.726525626075851e-06, - "loss": 0.0887, + "epoch": 1.53, + "learning_rate": 4.723456241670369e-06, + "loss": 0.054, "step": 13770 }, { - "epoch": 0.76, - "learning_rate": 4.720972846909879e-06, - "loss": 0.0785, + "epoch": 1.53, + "learning_rate": 4.71790315415371e-06, + "loss": 0.0703, "step": 13775 }, { - "epoch": 0.76, - "learning_rate": 4.715420067743906e-06, - "loss": 0.0925, + "epoch": 1.53, + "learning_rate": 4.712350066637051e-06, + "loss": 0.0677, "step": 13780 }, { - "epoch": 0.77, - "learning_rate": 4.7098672885779335e-06, - "loss": 0.0989, + "epoch": 1.53, + "learning_rate": 4.706796979120392e-06, + "loss": 0.0437, "step": 13785 }, { - "epoch": 0.77, - "learning_rate": 4.704314509411961e-06, - "loss": 0.1323, + "epoch": 1.53, + "learning_rate": 4.701243891603732e-06, + "loss": 0.0208, "step": 13790 }, { - "epoch": 0.77, - "learning_rate": 4.698761730245989e-06, - "loss": 0.1202, + "epoch": 1.53, + "learning_rate": 4.695690804087073e-06, + "loss": 0.0511, "step": 13795 }, { - "epoch": 0.77, - "learning_rate": 4.693208951080016e-06, - "loss": 0.2338, + "epoch": 1.53, + "learning_rate": 4.690137716570413e-06, + "loss": 0.0624, "step": 13800 }, { - "epoch": 0.77, - "learning_rate": 4.687656171914044e-06, - "loss": 0.0796, + "epoch": 1.53, + "learning_rate": 4.6845846290537545e-06, + "loss": 0.1252, "step": 13805 }, { - "epoch": 0.77, - "learning_rate": 4.682103392748071e-06, - "loss": 0.1078, + "epoch": 1.53, + "learning_rate": 4.6790315415370955e-06, + "loss": 0.2626, "step": 13810 }, { - "epoch": 0.77, - "learning_rate": 4.6765506135820985e-06, - "loss": 0.0927, + "epoch": 1.53, + "learning_rate": 4.673478454020436e-06, + "loss": 0.0563, "step": 13815 }, { - "epoch": 0.77, - "learning_rate": 4.6709978344161255e-06, - "loss": 0.1071, + "epoch": 1.53, + "learning_rate": 4.667925366503777e-06, + "loss": 0.0615, "step": 13820 }, { - "epoch": 0.77, - "learning_rate": 4.665445055250153e-06, - "loss": 0.0593, + "epoch": 1.53, + "learning_rate": 4.662372278987117e-06, + "loss": 0.0605, "step": 13825 }, { - "epoch": 0.77, - "learning_rate": 4.65989227608418e-06, - "loss": 0.0868, + "epoch": 1.54, + "learning_rate": 4.656819191470458e-06, + "loss": 0.0981, "step": 13830 }, { - "epoch": 0.77, - "learning_rate": 4.654339496918208e-06, - "loss": 0.0558, + "epoch": 1.54, + "learning_rate": 4.651266103953799e-06, + "loss": 0.0391, "step": 13835 }, { - "epoch": 0.77, - "learning_rate": 4.648786717752236e-06, - "loss": 0.1034, + "epoch": 1.54, + "learning_rate": 4.645713016437139e-06, + "loss": 0.0593, "step": 13840 }, { - "epoch": 0.77, - "learning_rate": 4.643233938586263e-06, - "loss": 0.1599, + "epoch": 1.54, + "learning_rate": 4.64015992892048e-06, + "loss": 0.0835, "step": 13845 }, { - "epoch": 0.77, - "learning_rate": 4.637681159420291e-06, - "loss": 0.2171, + "epoch": 1.54, + "learning_rate": 4.634606841403821e-06, + "loss": 0.0966, "step": 13850 }, { - "epoch": 0.77, - "learning_rate": 4.6321283802543175e-06, - "loss": 0.0787, + "epoch": 1.54, + "learning_rate": 4.629053753887162e-06, + "loss": 0.1804, "step": 13855 }, { - "epoch": 0.77, - "learning_rate": 4.626575601088345e-06, - "loss": 0.0754, + "epoch": 1.54, + "learning_rate": 4.623500666370502e-06, + "loss": 0.2938, "step": 13860 }, { - "epoch": 0.77, - "learning_rate": 4.621022821922372e-06, - "loss": 0.0757, + "epoch": 1.54, + "learning_rate": 4.617947578853843e-06, + "loss": 0.061, "step": 13865 }, { - "epoch": 0.77, - "learning_rate": 4.6154700427564e-06, - "loss": 0.128, + "epoch": 1.54, + "learning_rate": 4.612394491337184e-06, + "loss": 0.0798, "step": 13870 }, { - "epoch": 0.77, - "learning_rate": 4.609917263590427e-06, - "loss": 0.0651, + "epoch": 1.54, + "learning_rate": 4.606841403820524e-06, + "loss": 0.0693, "step": 13875 }, { - "epoch": 0.77, - "learning_rate": 4.604364484424455e-06, - "loss": 0.0779, + "epoch": 1.54, + "learning_rate": 4.601288316303865e-06, + "loss": 0.0772, "step": 13880 }, { - "epoch": 0.77, - "learning_rate": 4.598811705258483e-06, - "loss": 0.0598, + "epoch": 1.54, + "learning_rate": 4.5957352287872055e-06, + "loss": 0.0576, "step": 13885 }, { - "epoch": 0.77, - "learning_rate": 4.5932589260925096e-06, - "loss": 0.1017, + "epoch": 1.54, + "learning_rate": 4.590182141270547e-06, + "loss": 0.0567, "step": 13890 }, { - "epoch": 0.77, - "learning_rate": 4.587706146926537e-06, - "loss": 0.1389, + "epoch": 1.54, + "learning_rate": 4.584629053753888e-06, + "loss": 0.0605, "step": 13895 }, { - "epoch": 0.77, - "learning_rate": 4.582153367760564e-06, - "loss": 0.2635, + "epoch": 1.54, + "learning_rate": 4.579075966237228e-06, + "loss": 0.0796, "step": 13900 }, { - "epoch": 0.77, - "learning_rate": 4.576600588594592e-06, - "loss": 0.1182, + "epoch": 1.54, + "learning_rate": 4.573522878720569e-06, + "loss": 0.1112, "step": 13905 }, { - "epoch": 0.77, - "learning_rate": 4.571047809428619e-06, - "loss": 0.0883, + "epoch": 1.54, + "learning_rate": 4.567969791203909e-06, + "loss": 0.248, "step": 13910 }, { - "epoch": 0.77, - "learning_rate": 4.565495030262647e-06, - "loss": 0.1146, + "epoch": 1.54, + "learning_rate": 4.562416703687251e-06, + "loss": 0.0721, "step": 13915 }, { - "epoch": 0.77, - "learning_rate": 4.559942251096674e-06, - "loss": 0.1247, + "epoch": 1.55, + "learning_rate": 4.556863616170591e-06, + "loss": 0.0514, "step": 13920 }, { - "epoch": 0.77, - "learning_rate": 4.554389471930702e-06, - "loss": 0.1342, + "epoch": 1.55, + "learning_rate": 4.551310528653932e-06, + "loss": 0.0729, "step": 13925 }, { - "epoch": 0.77, - "learning_rate": 4.548836692764729e-06, - "loss": 0.0234, + "epoch": 1.55, + "learning_rate": 4.5457574411372725e-06, + "loss": 0.1085, "step": 13930 }, { - "epoch": 0.77, - "learning_rate": 4.543283913598756e-06, - "loss": 0.0556, + "epoch": 1.55, + "learning_rate": 4.540204353620614e-06, + "loss": 0.0554, "step": 13935 }, { - "epoch": 0.77, - "learning_rate": 4.537731134432784e-06, - "loss": 0.0993, + "epoch": 1.55, + "learning_rate": 4.534651266103955e-06, + "loss": 0.0583, "step": 13940 }, { - "epoch": 0.77, - "learning_rate": 4.532178355266811e-06, - "loss": 0.1551, + "epoch": 1.55, + "learning_rate": 4.529098178587295e-06, + "loss": 0.0531, "step": 13945 }, { - "epoch": 0.77, - "learning_rate": 4.526625576100839e-06, - "loss": 0.3049, + "epoch": 1.55, + "learning_rate": 4.523545091070636e-06, + "loss": 0.1009, "step": 13950 }, { - "epoch": 0.77, - "learning_rate": 4.521072796934866e-06, - "loss": 0.0981, + "epoch": 1.55, + "learning_rate": 4.517992003553976e-06, + "loss": 0.1851, "step": 13955 }, { - "epoch": 0.77, - "learning_rate": 4.515520017768894e-06, - "loss": 0.1149, + "epoch": 1.55, + "learning_rate": 4.512438916037317e-06, + "loss": 0.2653, "step": 13960 }, { - "epoch": 0.77, - "learning_rate": 4.509967238602921e-06, - "loss": 0.1163, + "epoch": 1.55, + "learning_rate": 4.506885828520658e-06, + "loss": 0.048, "step": 13965 }, { - "epoch": 0.78, - "learning_rate": 4.504414459436948e-06, - "loss": 0.1474, + "epoch": 1.55, + "learning_rate": 4.5013327410039985e-06, + "loss": 0.0815, "step": 13970 }, { - "epoch": 0.78, - "learning_rate": 4.498861680270976e-06, - "loss": 0.1149, + "epoch": 1.55, + "learning_rate": 4.4957796534873396e-06, + "loss": 0.0778, "step": 13975 }, { - "epoch": 0.78, - "learning_rate": 4.493308901105003e-06, - "loss": 0.0517, + "epoch": 1.55, + "learning_rate": 4.49022656597068e-06, + "loss": 0.0635, "step": 13980 }, { - "epoch": 0.78, - "learning_rate": 4.487756121939031e-06, - "loss": 0.0745, + "epoch": 1.55, + "learning_rate": 4.484673478454021e-06, + "loss": 0.0347, "step": 13985 }, { - "epoch": 0.78, - "learning_rate": 4.482203342773058e-06, - "loss": 0.1147, + "epoch": 1.55, + "learning_rate": 4.479120390937362e-06, + "loss": 0.0572, "step": 13990 }, { - "epoch": 0.78, - "learning_rate": 4.476650563607086e-06, - "loss": 0.1251, + "epoch": 1.55, + "learning_rate": 4.473567303420702e-06, + "loss": 0.0781, "step": 13995 }, { - "epoch": 0.78, - "learning_rate": 4.471097784441113e-06, - "loss": 0.1945, + "epoch": 1.55, + "learning_rate": 4.468014215904043e-06, + "loss": 0.1064, "step": 14000 }, { - "epoch": 0.78, - "eval_loss": 0.04153085872530937, - "eval_runtime": 1528.4265, - "eval_samples_per_second": 10.48, - "eval_steps_per_second": 2.62, - "eval_wer": 0.11186440677966102, + "epoch": 1.55, + "eval_loss": 0.02988557517528534, + "eval_runtime": 1484.3586, + "eval_samples_per_second": 10.791, + "eval_steps_per_second": 2.698, + "eval_wer": 0.0711503786512802, "step": 14000 }, { - "epoch": 0.78, - "learning_rate": 4.46554500527514e-06, - "loss": 0.1088, + "epoch": 1.55, + "learning_rate": 4.462461128387383e-06, + "loss": 0.1274, "step": 14005 }, { - "epoch": 0.78, - "learning_rate": 4.459992226109168e-06, - "loss": 0.081, + "epoch": 1.56, + "learning_rate": 4.4569080408707245e-06, + "loss": 0.274, "step": 14010 }, { - "epoch": 0.78, - "learning_rate": 4.454439446943196e-06, - "loss": 0.0944, + "epoch": 1.56, + "learning_rate": 4.451354953354065e-06, + "loss": 0.0532, "step": 14015 }, { - "epoch": 0.78, - "learning_rate": 4.448886667777223e-06, - "loss": 0.1004, + "epoch": 1.56, + "learning_rate": 4.445801865837406e-06, + "loss": 0.0759, "step": 14020 }, { - "epoch": 0.78, - "learning_rate": 4.443333888611251e-06, - "loss": 0.0912, + "epoch": 1.56, + "learning_rate": 4.440248778320747e-06, + "loss": 0.0842, "step": 14025 }, { - "epoch": 0.78, - "learning_rate": 4.437781109445278e-06, - "loss": 0.0567, + "epoch": 1.56, + "learning_rate": 4.434695690804087e-06, + "loss": 0.0861, "step": 14030 }, { - "epoch": 0.78, - "learning_rate": 4.4322283302793055e-06, - "loss": 0.1223, + "epoch": 1.56, + "learning_rate": 4.429142603287428e-06, + "loss": 0.0383, "step": 14035 }, { - "epoch": 0.78, - "learning_rate": 4.426675551113332e-06, - "loss": 0.1591, + "epoch": 1.56, + "learning_rate": 4.423589515770768e-06, + "loss": 0.0522, "step": 14040 }, { - "epoch": 0.78, - "learning_rate": 4.42112277194736e-06, - "loss": 0.1247, + "epoch": 1.56, + "learning_rate": 4.418036428254109e-06, + "loss": 0.0702, "step": 14045 }, { - "epoch": 0.78, - "learning_rate": 4.415569992781387e-06, - "loss": 0.3555, + "epoch": 1.56, + "learning_rate": 4.41248334073745e-06, + "loss": 0.0783, "step": 14050 }, { - "epoch": 0.78, - "learning_rate": 4.410017213615415e-06, - "loss": 0.1043, + "epoch": 1.56, + "learning_rate": 4.4069302532207915e-06, + "loss": 0.1226, "step": 14055 }, { - "epoch": 0.78, - "learning_rate": 4.404464434449443e-06, - "loss": 0.1261, + "epoch": 1.56, + "learning_rate": 4.401377165704132e-06, + "loss": 0.3258, "step": 14060 }, { - "epoch": 0.78, - "learning_rate": 4.39891165528347e-06, - "loss": 0.0907, + "epoch": 1.56, + "learning_rate": 4.395824078187473e-06, + "loss": 0.0789, "step": 14065 }, { - "epoch": 0.78, - "learning_rate": 4.3933588761174975e-06, - "loss": 0.1144, + "epoch": 1.56, + "learning_rate": 4.390270990670814e-06, + "loss": 0.0608, "step": 14070 }, { - "epoch": 0.78, - "learning_rate": 4.3878060969515244e-06, - "loss": 0.0496, + "epoch": 1.56, + "learning_rate": 4.384717903154154e-06, + "loss": 0.0651, "step": 14075 }, { - "epoch": 0.78, - "learning_rate": 4.382253317785552e-06, - "loss": 0.1049, + "epoch": 1.56, + "learning_rate": 4.379164815637495e-06, + "loss": 0.0817, "step": 14080 }, { - "epoch": 0.78, - "learning_rate": 4.376700538619579e-06, - "loss": 0.1369, + "epoch": 1.56, + "learning_rate": 4.373611728120835e-06, + "loss": 0.0539, "step": 14085 }, { - "epoch": 0.78, - "learning_rate": 4.371147759453607e-06, - "loss": 0.1644, + "epoch": 1.56, + "learning_rate": 4.368058640604176e-06, + "loss": 0.0422, "step": 14090 }, { - "epoch": 0.78, - "learning_rate": 4.365594980287634e-06, - "loss": 0.1771, + "epoch": 1.56, + "learning_rate": 4.3625055530875174e-06, + "loss": 0.0508, "step": 14095 }, { - "epoch": 0.78, - "learning_rate": 4.360042201121662e-06, - "loss": 0.2205, + "epoch": 1.57, + "learning_rate": 4.356952465570858e-06, + "loss": 0.1083, "step": 14100 }, { - "epoch": 0.78, - "learning_rate": 4.3544894219556895e-06, - "loss": 0.1173, + "epoch": 1.57, + "learning_rate": 4.351399378054199e-06, + "loss": 0.136, "step": 14105 }, { - "epoch": 0.78, - "learning_rate": 4.3489366427897165e-06, - "loss": 0.1243, + "epoch": 1.57, + "learning_rate": 4.345846290537539e-06, + "loss": 0.2332, "step": 14110 }, { - "epoch": 0.78, - "learning_rate": 4.343383863623744e-06, - "loss": 0.0949, + "epoch": 1.57, + "learning_rate": 4.34029320302088e-06, + "loss": 0.0539, "step": 14115 }, { - "epoch": 0.78, - "learning_rate": 4.337831084457771e-06, - "loss": 0.1238, + "epoch": 1.57, + "learning_rate": 4.334740115504221e-06, + "loss": 0.0642, "step": 14120 }, { - "epoch": 0.78, - "learning_rate": 4.332278305291799e-06, - "loss": 0.1289, + "epoch": 1.57, + "learning_rate": 4.329187027987561e-06, + "loss": 0.0592, "step": 14125 }, { - "epoch": 0.78, - "learning_rate": 4.326725526125826e-06, - "loss": 0.0113, + "epoch": 1.57, + "learning_rate": 4.323633940470902e-06, + "loss": 0.0759, "step": 14130 }, { - "epoch": 0.78, - "learning_rate": 4.321172746959854e-06, - "loss": 0.0867, + "epoch": 1.57, + "learning_rate": 4.3180808529542425e-06, + "loss": 0.0538, "step": 14135 }, { - "epoch": 0.78, - "learning_rate": 4.315619967793881e-06, - "loss": 0.1616, + "epoch": 1.57, + "learning_rate": 4.312527765437584e-06, + "loss": 0.0352, "step": 14140 }, { - "epoch": 0.78, - "learning_rate": 4.3100671886279085e-06, - "loss": 0.1899, + "epoch": 1.57, + "learning_rate": 4.306974677920925e-06, + "loss": 0.0949, "step": 14145 }, { - "epoch": 0.79, - "learning_rate": 4.304514409461936e-06, - "loss": 0.2842, + "epoch": 1.57, + "learning_rate": 4.301421590404265e-06, + "loss": 0.076, "step": 14150 }, { - "epoch": 0.79, - "learning_rate": 4.298961630295963e-06, - "loss": 0.0779, + "epoch": 1.57, + "learning_rate": 4.295868502887606e-06, + "loss": 0.1482, "step": 14155 }, { - "epoch": 0.79, - "learning_rate": 4.293408851129991e-06, - "loss": 0.0941, + "epoch": 1.57, + "learning_rate": 4.290315415370946e-06, + "loss": 0.3044, "step": 14160 }, { - "epoch": 0.79, - "learning_rate": 4.287856071964018e-06, - "loss": 0.1026, + "epoch": 1.57, + "learning_rate": 4.284762327854287e-06, + "loss": 0.0597, "step": 14165 }, { - "epoch": 0.79, - "learning_rate": 4.282303292798046e-06, - "loss": 0.1587, + "epoch": 1.57, + "learning_rate": 4.2792092403376274e-06, + "loss": 0.0785, "step": 14170 }, { - "epoch": 0.79, - "learning_rate": 4.276750513632073e-06, - "loss": 0.046, + "epoch": 1.57, + "learning_rate": 4.2736561528209685e-06, + "loss": 0.0642, "step": 14175 }, { - "epoch": 0.79, - "learning_rate": 4.2711977344661005e-06, - "loss": 0.0583, + "epoch": 1.57, + "learning_rate": 4.2681030653043096e-06, + "loss": 0.0689, "step": 14180 }, { - "epoch": 0.79, - "learning_rate": 4.2656449553001275e-06, - "loss": 0.0684, + "epoch": 1.57, + "learning_rate": 4.26254997778765e-06, + "loss": 0.0494, "step": 14185 }, { - "epoch": 0.79, - "learning_rate": 4.260092176134156e-06, - "loss": 0.2209, + "epoch": 1.58, + "learning_rate": 4.256996890270992e-06, + "loss": 0.0384, "step": 14190 }, { - "epoch": 0.79, - "learning_rate": 4.254539396968183e-06, - "loss": 0.1618, + "epoch": 1.58, + "learning_rate": 4.251443802754332e-06, + "loss": 0.0707, "step": 14195 }, { - "epoch": 0.79, - "learning_rate": 4.24898661780221e-06, - "loss": 0.2359, + "epoch": 1.58, + "learning_rate": 4.245890715237673e-06, + "loss": 0.1587, "step": 14200 }, { - "epoch": 0.79, - "learning_rate": 4.243433838636238e-06, - "loss": 0.0823, + "epoch": 1.58, + "learning_rate": 4.240337627721013e-06, + "loss": 0.1377, "step": 14205 }, { - "epoch": 0.79, - "learning_rate": 4.237881059470265e-06, - "loss": 0.1005, + "epoch": 1.58, + "learning_rate": 4.234784540204354e-06, + "loss": 0.4014, "step": 14210 }, { - "epoch": 0.79, - "learning_rate": 4.2323282803042926e-06, - "loss": 0.1159, + "epoch": 1.58, + "learning_rate": 4.2292314526876945e-06, + "loss": 0.0525, "step": 14215 }, { - "epoch": 0.79, - "learning_rate": 4.2267755011383195e-06, - "loss": 0.1163, + "epoch": 1.58, + "learning_rate": 4.2236783651710355e-06, + "loss": 0.0776, "step": 14220 }, { - "epoch": 0.79, - "learning_rate": 4.221222721972347e-06, - "loss": 0.1128, + "epoch": 1.58, + "learning_rate": 4.218125277654377e-06, + "loss": 0.0634, "step": 14225 }, { - "epoch": 0.79, - "learning_rate": 4.215669942806375e-06, - "loss": 0.0508, + "epoch": 1.58, + "learning_rate": 4.212572190137717e-06, + "loss": 0.0871, "step": 14230 }, { - "epoch": 0.79, - "learning_rate": 4.210117163640403e-06, - "loss": 0.0895, + "epoch": 1.58, + "learning_rate": 4.207019102621058e-06, + "loss": 0.0475, "step": 14235 }, { - "epoch": 0.79, - "learning_rate": 4.20456438447443e-06, - "loss": 0.0561, + "epoch": 1.58, + "learning_rate": 4.201466015104398e-06, + "loss": 0.0565, "step": 14240 }, { - "epoch": 0.79, - "learning_rate": 4.199011605308458e-06, - "loss": 0.1555, + "epoch": 1.58, + "learning_rate": 4.195912927587739e-06, + "loss": 0.0538, "step": 14245 }, { - "epoch": 0.79, - "learning_rate": 4.193458826142485e-06, - "loss": 0.202, + "epoch": 1.58, + "learning_rate": 4.19035984007108e-06, + "loss": 0.0968, "step": 14250 }, { - "epoch": 0.79, - "learning_rate": 4.187906046976512e-06, - "loss": 0.085, + "epoch": 1.58, + "learning_rate": 4.18480675255442e-06, + "loss": 0.1581, "step": 14255 }, { - "epoch": 0.79, - "learning_rate": 4.182353267810539e-06, - "loss": 0.1274, + "epoch": 1.58, + "learning_rate": 4.1792536650377615e-06, + "loss": 0.2344, "step": 14260 }, { - "epoch": 0.79, - "learning_rate": 4.176800488644567e-06, - "loss": 0.0962, + "epoch": 1.58, + "learning_rate": 4.173700577521102e-06, + "loss": 0.0859, "step": 14265 }, { - "epoch": 0.79, - "learning_rate": 4.171247709478594e-06, - "loss": 0.1147, + "epoch": 1.58, + "learning_rate": 4.168147490004443e-06, + "loss": 0.0623, "step": 14270 }, { - "epoch": 0.79, - "learning_rate": 4.165694930312622e-06, - "loss": 0.0231, + "epoch": 1.58, + "learning_rate": 4.162594402487784e-06, + "loss": 0.0936, "step": 14275 }, { - "epoch": 0.79, - "learning_rate": 4.16014215114665e-06, - "loss": 0.0666, + "epoch": 1.59, + "learning_rate": 4.157041314971124e-06, + "loss": 0.0774, "step": 14280 }, { - "epoch": 0.79, - "learning_rate": 4.154589371980677e-06, - "loss": 0.0807, + "epoch": 1.59, + "learning_rate": 4.151488227454465e-06, + "loss": 0.0648, "step": 14285 }, { - "epoch": 0.79, - "learning_rate": 4.149036592814704e-06, - "loss": 0.1702, + "epoch": 1.59, + "learning_rate": 4.145935139937805e-06, + "loss": 0.0513, "step": 14290 }, { - "epoch": 0.79, - "learning_rate": 4.143483813648731e-06, - "loss": 0.1171, + "epoch": 1.59, + "learning_rate": 4.140382052421146e-06, + "loss": 0.0832, "step": 14295 }, { - "epoch": 0.79, - "learning_rate": 4.137931034482759e-06, - "loss": 0.4418, + "epoch": 1.59, + "learning_rate": 4.1348289649044874e-06, + "loss": 0.1355, "step": 14300 }, { - "epoch": 0.79, - "learning_rate": 4.132378255316786e-06, - "loss": 0.1599, + "epoch": 1.59, + "learning_rate": 4.129275877387828e-06, + "loss": 0.1456, "step": 14305 }, { - "epoch": 0.79, - "learning_rate": 4.126825476150814e-06, - "loss": 0.1029, + "epoch": 1.59, + "learning_rate": 4.123722789871169e-06, + "loss": 0.2033, "step": 14310 }, { - "epoch": 0.79, - "learning_rate": 4.121272696984841e-06, - "loss": 0.1243, + "epoch": 1.59, + "learning_rate": 4.118169702354509e-06, + "loss": 0.0619, "step": 14315 }, { - "epoch": 0.79, - "learning_rate": 4.115719917818869e-06, - "loss": 0.1071, + "epoch": 1.59, + "learning_rate": 4.11261661483785e-06, + "loss": 0.0808, "step": 14320 }, { - "epoch": 0.79, - "learning_rate": 4.1101671386528964e-06, - "loss": 0.0725, + "epoch": 1.59, + "learning_rate": 4.107063527321191e-06, + "loss": 0.0802, "step": 14325 }, { - "epoch": 0.8, - "learning_rate": 4.104614359486923e-06, - "loss": 0.0992, + "epoch": 1.59, + "learning_rate": 4.101510439804532e-06, + "loss": 0.076, "step": 14330 }, { - "epoch": 0.8, - "learning_rate": 4.099061580320951e-06, - "loss": 0.1229, + "epoch": 1.59, + "learning_rate": 4.095957352287872e-06, + "loss": 0.0421, "step": 14335 }, { - "epoch": 0.8, - "learning_rate": 4.093508801154978e-06, - "loss": 0.0982, + "epoch": 1.59, + "learning_rate": 4.090404264771213e-06, + "loss": 0.0628, "step": 14340 }, { - "epoch": 0.8, - "learning_rate": 4.087956021989006e-06, - "loss": 0.1309, + "epoch": 1.59, + "learning_rate": 4.0848511772545545e-06, + "loss": 0.0431, "step": 14345 }, { - "epoch": 0.8, - "learning_rate": 4.082403242823033e-06, - "loss": 0.3549, + "epoch": 1.59, + "learning_rate": 4.079298089737895e-06, + "loss": 0.0715, "step": 14350 }, { - "epoch": 0.8, - "learning_rate": 4.076850463657061e-06, - "loss": 0.0913, + "epoch": 1.59, + "learning_rate": 4.073745002221236e-06, + "loss": 0.1872, "step": 14355 }, { - "epoch": 0.8, - "learning_rate": 4.071297684491088e-06, - "loss": 0.122, + "epoch": 1.59, + "learning_rate": 4.068191914704576e-06, + "loss": 0.1892, "step": 14360 }, { - "epoch": 0.8, - "learning_rate": 4.0657449053251154e-06, - "loss": 0.1, + "epoch": 1.59, + "learning_rate": 4.062638827187917e-06, + "loss": 0.0398, "step": 14365 }, { - "epoch": 0.8, - "learning_rate": 4.060192126159143e-06, - "loss": 0.1071, + "epoch": 1.6, + "learning_rate": 4.057085739671257e-06, + "loss": 0.0525, "step": 14370 }, { - "epoch": 0.8, - "learning_rate": 4.05463934699317e-06, - "loss": 0.1482, + "epoch": 1.6, + "learning_rate": 4.051532652154598e-06, + "loss": 0.0606, "step": 14375 }, { - "epoch": 0.8, - "learning_rate": 4.049086567827198e-06, - "loss": 0.0353, + "epoch": 1.6, + "learning_rate": 4.045979564637939e-06, + "loss": 0.0792, "step": 14380 }, { - "epoch": 0.8, - "learning_rate": 4.043533788661225e-06, - "loss": 0.0894, + "epoch": 1.6, + "learning_rate": 4.0404264771212796e-06, + "loss": 0.0572, "step": 14385 }, { - "epoch": 0.8, - "learning_rate": 4.037981009495253e-06, - "loss": 0.2058, + "epoch": 1.6, + "learning_rate": 4.034873389604621e-06, + "loss": 0.0396, "step": 14390 }, { - "epoch": 0.8, - "learning_rate": 4.03242823032928e-06, - "loss": 0.1188, + "epoch": 1.6, + "learning_rate": 4.029320302087961e-06, + "loss": 0.0778, "step": 14395 }, { - "epoch": 0.8, - "learning_rate": 4.0268754511633075e-06, - "loss": 0.2822, + "epoch": 1.6, + "learning_rate": 4.023767214571302e-06, + "loss": 0.1087, "step": 14400 }, { - "epoch": 0.8, - "learning_rate": 4.021322671997334e-06, - "loss": 0.0769, + "epoch": 1.6, + "learning_rate": 4.018214127054643e-06, + "loss": 0.1612, "step": 14405 }, { - "epoch": 0.8, - "learning_rate": 4.015769892831363e-06, - "loss": 0.0912, + "epoch": 1.6, + "learning_rate": 4.012661039537983e-06, + "loss": 0.1456, "step": 14410 }, { - "epoch": 0.8, - "learning_rate": 4.01021711366539e-06, - "loss": 0.1024, + "epoch": 1.6, + "learning_rate": 4.007107952021324e-06, + "loss": 0.0603, "step": 14415 }, { - "epoch": 0.8, - "learning_rate": 4.004664334499418e-06, - "loss": 0.1484, + "epoch": 1.6, + "learning_rate": 4.0015548645046645e-06, + "loss": 0.0838, "step": 14420 }, { - "epoch": 0.8, - "learning_rate": 3.999111555333445e-06, - "loss": 0.1103, + "epoch": 1.6, + "learning_rate": 3.9960017769880055e-06, + "loss": 0.0697, "step": 14425 }, { - "epoch": 0.8, - "learning_rate": 3.9935587761674725e-06, - "loss": 0.0498, + "epoch": 1.6, + "learning_rate": 3.990448689471347e-06, + "loss": 0.0712, "step": 14430 }, { - "epoch": 0.8, - "learning_rate": 3.9880059970014995e-06, - "loss": 0.0451, + "epoch": 1.6, + "learning_rate": 3.984895601954687e-06, + "loss": 0.0524, "step": 14435 }, { - "epoch": 0.8, - "learning_rate": 3.9824532178355264e-06, - "loss": 0.0829, + "epoch": 1.6, + "learning_rate": 3.979342514438028e-06, + "loss": 0.0306, "step": 14440 }, { - "epoch": 0.8, - "learning_rate": 3.976900438669554e-06, - "loss": 0.141, + "epoch": 1.6, + "learning_rate": 3.973789426921368e-06, + "loss": 0.0625, "step": 14445 }, { - "epoch": 0.8, - "learning_rate": 3.971347659503582e-06, - "loss": 0.2178, + "epoch": 1.6, + "learning_rate": 3.968236339404709e-06, + "loss": 0.0956, "step": 14450 }, { - "epoch": 0.8, - "learning_rate": 3.96579488033761e-06, - "loss": 0.0747, + "epoch": 1.6, + "learning_rate": 3.96268325188805e-06, + "loss": 0.185, "step": 14455 }, { - "epoch": 0.8, - "learning_rate": 3.960242101171637e-06, - "loss": 0.0957, + "epoch": 1.61, + "learning_rate": 3.957130164371391e-06, + "loss": 0.246, "step": 14460 }, { - "epoch": 0.8, - "learning_rate": 3.9546893220056646e-06, - "loss": 0.1182, + "epoch": 1.61, + "learning_rate": 3.9515770768547315e-06, + "loss": 0.0717, "step": 14465 }, { - "epoch": 0.8, - "learning_rate": 3.9491365428396915e-06, - "loss": 0.0885, + "epoch": 1.61, + "learning_rate": 3.9460239893380725e-06, + "loss": 0.0477, "step": 14470 }, { - "epoch": 0.8, - "learning_rate": 3.943583763673719e-06, - "loss": 0.1203, + "epoch": 1.61, + "learning_rate": 3.940470901821414e-06, + "loss": 0.0831, "step": 14475 }, { - "epoch": 0.8, - "learning_rate": 3.938030984507746e-06, - "loss": 0.06, + "epoch": 1.61, + "learning_rate": 3.934917814304754e-06, + "loss": 0.0942, "step": 14480 }, { - "epoch": 0.8, - "learning_rate": 3.932478205341774e-06, - "loss": 0.1195, + "epoch": 1.61, + "learning_rate": 3.929364726788095e-06, + "loss": 0.069, "step": 14485 }, { - "epoch": 0.8, - "learning_rate": 3.926925426175801e-06, - "loss": 0.1868, + "epoch": 1.61, + "learning_rate": 3.923811639271435e-06, + "loss": 0.036, "step": 14490 }, { - "epoch": 0.8, - "learning_rate": 3.921372647009829e-06, - "loss": 0.1726, + "epoch": 1.61, + "learning_rate": 3.918258551754776e-06, + "loss": 0.0794, "step": 14495 }, { - "epoch": 0.8, - "learning_rate": 3.915819867843857e-06, - "loss": 0.2437, + "epoch": 1.61, + "learning_rate": 3.912705464238117e-06, + "loss": 0.0835, "step": 14500 }, { - "epoch": 0.8, - "learning_rate": 3.9102670886778835e-06, - "loss": 0.1034, + "epoch": 1.61, + "learning_rate": 3.9071523767214574e-06, + "loss": 0.1196, "step": 14505 }, { - "epoch": 0.81, - "learning_rate": 3.904714309511911e-06, - "loss": 0.1042, + "epoch": 1.61, + "learning_rate": 3.9015992892047985e-06, + "loss": 0.1931, "step": 14510 }, { - "epoch": 0.81, - "learning_rate": 3.899161530345938e-06, - "loss": 0.1125, + "epoch": 1.61, + "learning_rate": 3.896046201688139e-06, + "loss": 0.0439, "step": 14515 }, { - "epoch": 0.81, - "learning_rate": 3.893608751179966e-06, - "loss": 0.1234, + "epoch": 1.61, + "learning_rate": 3.89049311417148e-06, + "loss": 0.0606, "step": 14520 }, { - "epoch": 0.81, - "learning_rate": 3.888055972013993e-06, - "loss": 0.1099, + "epoch": 1.61, + "learning_rate": 3.88494002665482e-06, + "loss": 0.085, "step": 14525 }, { - "epoch": 0.81, - "learning_rate": 3.882503192848021e-06, - "loss": 0.0186, + "epoch": 1.61, + "learning_rate": 3.879386939138161e-06, + "loss": 0.062, "step": 14530 }, { - "epoch": 0.81, - "learning_rate": 3.876950413682048e-06, - "loss": 0.0503, + "epoch": 1.61, + "learning_rate": 3.873833851621502e-06, + "loss": 0.0491, "step": 14535 }, { - "epoch": 0.81, - "learning_rate": 3.8713976345160756e-06, - "loss": 0.0696, + "epoch": 1.61, + "learning_rate": 3.868280764104842e-06, + "loss": 0.0549, "step": 14540 }, { - "epoch": 0.81, - "learning_rate": 3.865844855350103e-06, - "loss": 0.2068, + "epoch": 1.61, + "learning_rate": 3.862727676588183e-06, + "loss": 0.0557, "step": 14545 }, { - "epoch": 0.81, - "learning_rate": 3.86029207618413e-06, - "loss": 0.289, + "epoch": 1.62, + "learning_rate": 3.857174589071524e-06, + "loss": 0.0499, "step": 14550 }, { - "epoch": 0.81, - "learning_rate": 3.854739297018158e-06, - "loss": 0.1037, + "epoch": 1.62, + "learning_rate": 3.851621501554865e-06, + "loss": 0.1405, "step": 14555 }, { - "epoch": 0.81, - "learning_rate": 3.849186517852185e-06, - "loss": 0.0798, + "epoch": 1.62, + "learning_rate": 3.846068414038206e-06, + "loss": 0.2988, "step": 14560 }, { - "epoch": 0.81, - "learning_rate": 3.843633738686213e-06, - "loss": 0.1461, + "epoch": 1.62, + "learning_rate": 3.840515326521546e-06, + "loss": 0.0564, "step": 14565 }, { - "epoch": 0.81, - "learning_rate": 3.83808095952024e-06, - "loss": 0.1341, + "epoch": 1.62, + "learning_rate": 3.834962239004887e-06, + "loss": 0.0673, "step": 14570 }, { - "epoch": 0.81, - "learning_rate": 3.832528180354268e-06, - "loss": 0.1317, + "epoch": 1.62, + "learning_rate": 3.829409151488227e-06, + "loss": 0.0862, "step": 14575 }, { - "epoch": 0.81, - "learning_rate": 3.8269754011882946e-06, - "loss": 0.0655, + "epoch": 1.62, + "learning_rate": 3.823856063971568e-06, + "loss": 0.0788, "step": 14580 }, { - "epoch": 0.81, - "learning_rate": 3.821422622022322e-06, - "loss": 0.0569, + "epoch": 1.62, + "learning_rate": 3.818302976454909e-06, + "loss": 0.029, "step": 14585 }, { - "epoch": 0.81, - "learning_rate": 3.81586984285635e-06, - "loss": 0.1779, + "epoch": 1.62, + "learning_rate": 3.8127498889382496e-06, + "loss": 0.041, "step": 14590 }, { - "epoch": 0.81, - "learning_rate": 3.8103170636903775e-06, - "loss": 0.094, + "epoch": 1.62, + "learning_rate": 3.807196801421591e-06, + "loss": 0.0894, "step": 14595 }, { - "epoch": 0.81, - "learning_rate": 3.804764284524405e-06, - "loss": 0.4238, + "epoch": 1.62, + "learning_rate": 3.8016437139049317e-06, + "loss": 0.0573, "step": 14600 }, { - "epoch": 0.81, - "learning_rate": 3.7992115053584323e-06, - "loss": 0.0895, + "epoch": 1.62, + "learning_rate": 3.7960906263882723e-06, + "loss": 0.1123, "step": 14605 }, { - "epoch": 0.81, - "learning_rate": 3.7936587261924596e-06, - "loss": 0.0698, + "epoch": 1.62, + "learning_rate": 3.790537538871613e-06, + "loss": 0.1958, "step": 14610 }, { - "epoch": 0.81, - "learning_rate": 3.788105947026487e-06, - "loss": 0.1134, + "epoch": 1.62, + "learning_rate": 3.784984451354954e-06, + "loss": 0.0716, "step": 14615 }, { - "epoch": 0.81, - "learning_rate": 3.7825531678605144e-06, - "loss": 0.1032, + "epoch": 1.62, + "learning_rate": 3.7794313638382947e-06, + "loss": 0.0662, "step": 14620 }, { - "epoch": 0.81, - "learning_rate": 3.7770003886945417e-06, - "loss": 0.0423, + "epoch": 1.62, + "learning_rate": 3.7738782763216353e-06, + "loss": 0.0947, "step": 14625 }, { - "epoch": 0.81, - "learning_rate": 3.7714476095285695e-06, - "loss": 0.1007, + "epoch": 1.62, + "learning_rate": 3.768325188804976e-06, + "loss": 0.0735, "step": 14630 }, { - "epoch": 0.81, - "learning_rate": 3.765894830362597e-06, - "loss": 0.0879, + "epoch": 1.62, + "learning_rate": 3.7627721012883166e-06, + "loss": 0.0457, "step": 14635 }, { - "epoch": 0.81, - "learning_rate": 3.7603420511966243e-06, - "loss": 0.0835, + "epoch": 1.62, + "learning_rate": 3.7572190137716572e-06, + "loss": 0.047, "step": 14640 }, { - "epoch": 0.81, - "learning_rate": 3.7547892720306517e-06, - "loss": 0.1928, + "epoch": 1.63, + "learning_rate": 3.7516659262549983e-06, + "loss": 0.0534, "step": 14645 }, { - "epoch": 0.81, - "learning_rate": 3.749236492864679e-06, - "loss": 0.4373, + "epoch": 1.63, + "learning_rate": 3.746112838738339e-06, + "loss": 0.1204, "step": 14650 }, { - "epoch": 0.81, - "learning_rate": 3.7436837136987064e-06, - "loss": 0.0636, + "epoch": 1.63, + "learning_rate": 3.7405597512216796e-06, + "loss": 0.1612, "step": 14655 }, { - "epoch": 0.81, - "learning_rate": 3.7381309345327338e-06, - "loss": 0.1357, + "epoch": 1.63, + "learning_rate": 3.73500666370502e-06, + "loss": 0.2189, "step": 14660 }, { - "epoch": 0.81, - "learning_rate": 3.732578155366761e-06, - "loss": 0.1509, + "epoch": 1.63, + "learning_rate": 3.729453576188361e-06, + "loss": 0.0668, "step": 14665 }, { - "epoch": 0.81, - "learning_rate": 3.7270253762007885e-06, - "loss": 0.1373, + "epoch": 1.63, + "learning_rate": 3.723900488671702e-06, + "loss": 0.047, "step": 14670 }, { - "epoch": 0.81, - "learning_rate": 3.7214725970348163e-06, - "loss": 0.1029, + "epoch": 1.63, + "learning_rate": 3.7183474011550425e-06, + "loss": 0.0492, "step": 14675 }, { - "epoch": 0.81, - "learning_rate": 3.7159198178688437e-06, - "loss": 0.0242, + "epoch": 1.63, + "learning_rate": 3.7139049311417153e-06, + "loss": 0.2949, "step": 14680 }, { - "epoch": 0.81, - "learning_rate": 3.710367038702871e-06, - "loss": 0.1826, + "epoch": 1.63, + "learning_rate": 3.708351843625056e-06, + "loss": 0.0647, "step": 14685 }, { - "epoch": 0.82, - "learning_rate": 3.7048142595368984e-06, - "loss": 0.1429, + "epoch": 1.63, + "learning_rate": 3.7027987561083966e-06, + "loss": 0.0396, "step": 14690 }, { - "epoch": 0.82, - "learning_rate": 3.699261480370926e-06, - "loss": 0.1313, + "epoch": 1.63, + "learning_rate": 3.6972456685917372e-06, + "loss": 0.0644, "step": 14695 }, { - "epoch": 0.82, - "learning_rate": 3.693708701204953e-06, - "loss": 0.3209, + "epoch": 1.63, + "learning_rate": 3.6916925810750783e-06, + "loss": 0.0838, "step": 14700 }, { - "epoch": 0.82, - "learning_rate": 3.6881559220389806e-06, - "loss": 0.0815, + "epoch": 1.63, + "learning_rate": 3.686139493558419e-06, + "loss": 0.1349, "step": 14705 }, { - "epoch": 0.82, - "learning_rate": 3.682603142873008e-06, - "loss": 0.0965, + "epoch": 1.63, + "learning_rate": 3.6805864060417596e-06, + "loss": 0.2417, "step": 14710 }, { - "epoch": 0.82, - "learning_rate": 3.677050363707036e-06, - "loss": 0.1116, + "epoch": 1.63, + "learning_rate": 3.6750333185251e-06, + "loss": 0.0607, "step": 14715 }, { - "epoch": 0.82, - "learning_rate": 3.6714975845410635e-06, - "loss": 0.1201, + "epoch": 1.63, + "learning_rate": 3.669480231008441e-06, + "loss": 0.109, "step": 14720 }, { - "epoch": 0.82, - "learning_rate": 3.665944805375091e-06, - "loss": 0.1063, + "epoch": 1.63, + "learning_rate": 3.663927143491782e-06, + "loss": 0.0655, "step": 14725 }, { - "epoch": 0.82, - "learning_rate": 3.660392026209118e-06, - "loss": 0.0726, + "epoch": 1.63, + "learning_rate": 3.6583740559751225e-06, + "loss": 0.0642, "step": 14730 }, { - "epoch": 0.82, - "learning_rate": 3.654839247043145e-06, - "loss": 0.087, + "epoch": 1.64, + "learning_rate": 3.652820968458463e-06, + "loss": 0.0312, "step": 14735 }, { - "epoch": 0.82, - "learning_rate": 3.6492864678771726e-06, - "loss": 0.1011, + "epoch": 1.64, + "learning_rate": 3.647267880941804e-06, + "loss": 0.0586, "step": 14740 }, { - "epoch": 0.82, - "learning_rate": 3.6437336887112e-06, - "loss": 0.1075, + "epoch": 1.64, + "learning_rate": 3.6417147934251445e-06, + "loss": 0.0513, "step": 14745 }, { - "epoch": 0.82, - "learning_rate": 3.6381809095452273e-06, - "loss": 0.2526, + "epoch": 1.64, + "learning_rate": 3.636161705908485e-06, + "loss": 0.0907, "step": 14750 }, { - "epoch": 0.82, - "learning_rate": 3.6326281303792547e-06, - "loss": 0.0782, + "epoch": 1.64, + "learning_rate": 3.630608618391826e-06, + "loss": 0.1115, "step": 14755 }, { - "epoch": 0.82, - "learning_rate": 3.627075351213283e-06, - "loss": 0.0912, + "epoch": 1.64, + "learning_rate": 3.625055530875167e-06, + "loss": 0.2793, "step": 14760 }, { - "epoch": 0.82, - "learning_rate": 3.6215225720473103e-06, - "loss": 0.131, + "epoch": 1.64, + "learning_rate": 3.6195024433585074e-06, + "loss": 0.0481, "step": 14765 }, { - "epoch": 0.82, - "learning_rate": 3.6159697928813377e-06, - "loss": 0.1226, + "epoch": 1.64, + "learning_rate": 3.613949355841848e-06, + "loss": 0.0487, "step": 14770 }, { - "epoch": 0.82, - "learning_rate": 3.610417013715365e-06, - "loss": 0.0811, + "epoch": 1.64, + "learning_rate": 3.6083962683251887e-06, + "loss": 0.07, "step": 14775 }, { - "epoch": 0.82, - "learning_rate": 3.6048642345493924e-06, - "loss": 0.0591, + "epoch": 1.64, + "learning_rate": 3.6028431808085298e-06, + "loss": 0.0945, "step": 14780 }, { - "epoch": 0.82, - "learning_rate": 3.5993114553834198e-06, - "loss": 0.066, + "epoch": 1.64, + "learning_rate": 3.597290093291871e-06, + "loss": 0.0244, "step": 14785 }, { - "epoch": 0.82, - "learning_rate": 3.593758676217447e-06, - "loss": 0.166, + "epoch": 1.64, + "learning_rate": 3.5917370057752115e-06, + "loss": 0.0691, "step": 14790 }, { - "epoch": 0.82, - "learning_rate": 3.5882058970514745e-06, - "loss": 0.2985, + "epoch": 1.64, + "learning_rate": 3.586183918258552e-06, + "loss": 0.0779, "step": 14795 }, { - "epoch": 0.82, - "learning_rate": 3.582653117885502e-06, - "loss": 0.2551, + "epoch": 1.64, + "learning_rate": 3.580630830741893e-06, + "loss": 0.1174, "step": 14800 }, { - "epoch": 0.82, - "learning_rate": 3.5771003387195297e-06, - "loss": 0.0706, + "epoch": 1.64, + "learning_rate": 3.575077743225234e-06, + "loss": 0.1204, "step": 14805 }, { - "epoch": 0.82, - "learning_rate": 3.571547559553557e-06, - "loss": 0.1029, + "epoch": 1.64, + "learning_rate": 3.5695246557085745e-06, + "loss": 0.3551, "step": 14810 }, { - "epoch": 0.82, - "learning_rate": 3.5659947803875844e-06, - "loss": 0.1348, + "epoch": 1.64, + "learning_rate": 3.563971568191915e-06, + "loss": 0.0683, "step": 14815 }, { - "epoch": 0.82, - "learning_rate": 3.560442001221612e-06, - "loss": 0.1338, + "epoch": 1.64, + "learning_rate": 3.5584184806752557e-06, + "loss": 0.0703, "step": 14820 }, { - "epoch": 0.82, - "learning_rate": 3.554889222055639e-06, - "loss": 0.0816, + "epoch": 1.65, + "learning_rate": 3.5528653931585964e-06, + "loss": 0.0708, "step": 14825 }, { - "epoch": 0.82, - "learning_rate": 3.5493364428896666e-06, - "loss": 0.0506, + "epoch": 1.65, + "learning_rate": 3.5473123056419374e-06, + "loss": 0.0737, "step": 14830 }, { - "epoch": 0.82, - "learning_rate": 3.543783663723694e-06, - "loss": 0.0892, + "epoch": 1.65, + "learning_rate": 3.541759218125278e-06, + "loss": 0.0544, "step": 14835 }, { - "epoch": 0.82, - "learning_rate": 3.5382308845577213e-06, - "loss": 0.1138, + "epoch": 1.65, + "learning_rate": 3.5362061306086187e-06, + "loss": 0.0621, "step": 14840 }, { - "epoch": 0.82, - "learning_rate": 3.5326781053917487e-06, - "loss": 0.168, + "epoch": 1.65, + "learning_rate": 3.5306530430919594e-06, + "loss": 0.0627, "step": 14845 }, { - "epoch": 0.82, - "learning_rate": 3.5271253262257765e-06, - "loss": 0.2979, + "epoch": 1.65, + "learning_rate": 3.5250999555753e-06, + "loss": 0.0836, "step": 14850 }, { - "epoch": 0.82, - "learning_rate": 3.521572547059804e-06, - "loss": 0.0913, + "epoch": 1.65, + "learning_rate": 3.519546868058641e-06, + "loss": 0.1198, "step": 14855 }, { - "epoch": 0.82, - "learning_rate": 3.516019767893831e-06, - "loss": 0.0858, + "epoch": 1.65, + "learning_rate": 3.5139937805419817e-06, + "loss": 0.198, "step": 14860 }, { - "epoch": 0.82, - "learning_rate": 3.5104669887278586e-06, - "loss": 0.1071, + "epoch": 1.65, + "learning_rate": 3.5084406930253223e-06, + "loss": 0.059, "step": 14865 }, { - "epoch": 0.83, - "learning_rate": 3.504914209561886e-06, - "loss": 0.109, + "epoch": 1.65, + "learning_rate": 3.502887605508663e-06, + "loss": 0.0931, "step": 14870 }, { - "epoch": 0.83, - "learning_rate": 3.4993614303959133e-06, - "loss": 0.0992, + "epoch": 1.65, + "learning_rate": 3.4973345179920036e-06, + "loss": 0.0586, "step": 14875 }, { - "epoch": 0.83, - "learning_rate": 3.4938086512299407e-06, - "loss": 0.0385, + "epoch": 1.65, + "learning_rate": 3.4917814304753447e-06, + "loss": 0.0669, "step": 14880 }, { - "epoch": 0.83, - "learning_rate": 3.488255872063968e-06, - "loss": 0.1536, + "epoch": 1.65, + "learning_rate": 3.4862283429586853e-06, + "loss": 0.085, "step": 14885 }, { - "epoch": 0.83, - "learning_rate": 3.4827030928979954e-06, - "loss": 0.1854, + "epoch": 1.65, + "learning_rate": 3.480675255442026e-06, + "loss": 0.0224, "step": 14890 }, { - "epoch": 0.83, - "learning_rate": 3.4771503137320232e-06, - "loss": 0.1021, + "epoch": 1.65, + "learning_rate": 3.4751221679253666e-06, + "loss": 0.0439, "step": 14895 }, { - "epoch": 0.83, - "learning_rate": 3.4715975345660506e-06, - "loss": 0.2254, + "epoch": 1.65, + "learning_rate": 3.4695690804087072e-06, + "loss": 0.1035, "step": 14900 }, { - "epoch": 0.83, - "learning_rate": 3.466044755400078e-06, - "loss": 0.0906, + "epoch": 1.65, + "learning_rate": 3.464015992892048e-06, + "loss": 0.1673, "step": 14905 }, { - "epoch": 0.83, - "learning_rate": 3.4604919762341054e-06, - "loss": 0.1112, + "epoch": 1.65, + "learning_rate": 3.458462905375389e-06, + "loss": 0.231, "step": 14910 }, { - "epoch": 0.83, - "learning_rate": 3.4549391970681327e-06, - "loss": 0.1181, + "epoch": 1.66, + "learning_rate": 3.4529098178587296e-06, + "loss": 0.0655, "step": 14915 }, { - "epoch": 0.83, - "learning_rate": 3.44938641790216e-06, - "loss": 0.1133, + "epoch": 1.66, + "learning_rate": 3.4473567303420706e-06, + "loss": 0.0727, "step": 14920 }, { - "epoch": 0.83, - "learning_rate": 3.4438336387361875e-06, - "loss": 0.0744, + "epoch": 1.66, + "learning_rate": 3.4418036428254113e-06, + "loss": 0.0828, "step": 14925 }, { - "epoch": 0.83, - "learning_rate": 3.438280859570215e-06, - "loss": 0.0784, + "epoch": 1.66, + "learning_rate": 3.4362505553087523e-06, + "loss": 0.1114, "step": 14930 }, { - "epoch": 0.83, - "learning_rate": 3.433838636237437e-06, - "loss": 0.5869, + "epoch": 1.66, + "learning_rate": 3.430697467792093e-06, + "loss": 0.0504, "step": 14935 }, { - "epoch": 0.83, - "learning_rate": 3.4282858570714643e-06, - "loss": 0.1209, + "epoch": 1.66, + "learning_rate": 3.4251443802754336e-06, + "loss": 0.0244, "step": 14940 }, { - "epoch": 0.83, - "learning_rate": 3.4227330779054917e-06, - "loss": 0.1873, + "epoch": 1.66, + "learning_rate": 3.4195912927587742e-06, + "loss": 0.0445, "step": 14945 }, { - "epoch": 0.83, - "learning_rate": 3.417180298739519e-06, - "loss": 0.2828, + "epoch": 1.66, + "learning_rate": 3.414038205242115e-06, + "loss": 0.0621, "step": 14950 }, { - "epoch": 0.83, - "learning_rate": 3.4116275195735472e-06, - "loss": 0.0688, + "epoch": 1.66, + "learning_rate": 3.408485117725456e-06, + "loss": 0.1206, "step": 14955 }, { - "epoch": 0.83, - "learning_rate": 3.4060747404075746e-06, - "loss": 0.1005, + "epoch": 1.66, + "learning_rate": 3.4029320302087966e-06, + "loss": 0.1914, "step": 14960 }, { - "epoch": 0.83, - "learning_rate": 3.400521961241602e-06, - "loss": 0.1435, + "epoch": 1.66, + "learning_rate": 3.3973789426921372e-06, + "loss": 0.0827, "step": 14965 }, { - "epoch": 0.83, - "learning_rate": 3.3949691820756294e-06, - "loss": 0.083, + "epoch": 1.66, + "learning_rate": 3.391825855175478e-06, + "loss": 0.0536, "step": 14970 }, { - "epoch": 0.83, - "learning_rate": 3.3894164029096567e-06, - "loss": 0.0462, + "epoch": 1.66, + "learning_rate": 3.3862727676588185e-06, + "loss": 0.0774, "step": 14975 }, { - "epoch": 0.83, - "learning_rate": 3.383863623743684e-06, - "loss": 0.0623, + "epoch": 1.66, + "learning_rate": 3.380719680142159e-06, + "loss": 0.0964, "step": 14980 }, { - "epoch": 0.83, - "learning_rate": 3.3783108445777115e-06, - "loss": 0.1096, + "epoch": 1.66, + "learning_rate": 3.3751665926255e-06, + "loss": 0.0797, "step": 14985 }, { - "epoch": 0.83, - "learning_rate": 3.372758065411739e-06, - "loss": 0.1172, + "epoch": 1.66, + "learning_rate": 3.369613505108841e-06, + "loss": 0.0568, "step": 14990 }, { - "epoch": 0.83, - "learning_rate": 3.3672052862457662e-06, - "loss": 0.2284, + "epoch": 1.66, + "learning_rate": 3.3640604175921815e-06, + "loss": 0.0553, "step": 14995 }, { - "epoch": 0.83, - "learning_rate": 3.361652507079794e-06, - "loss": 0.4134, + "epoch": 1.66, + "learning_rate": 3.358507330075522e-06, + "loss": 0.0645, "step": 15000 }, { - "epoch": 0.83, - "learning_rate": 3.3560997279138214e-06, - "loss": 0.0863, + "epoch": 1.67, + "learning_rate": 3.3529542425588628e-06, + "loss": 0.1871, "step": 15005 }, { - "epoch": 0.83, - "learning_rate": 3.3505469487478488e-06, - "loss": 0.093, + "epoch": 1.67, + "learning_rate": 3.347401155042204e-06, + "loss": 0.2664, "step": 15010 }, { - "epoch": 0.83, - "learning_rate": 3.344994169581876e-06, - "loss": 0.1008, + "epoch": 1.67, + "learning_rate": 3.3418480675255445e-06, + "loss": 0.0529, "step": 15015 }, { - "epoch": 0.83, - "learning_rate": 3.3394413904159035e-06, - "loss": 0.1121, + "epoch": 1.67, + "learning_rate": 3.336294980008885e-06, + "loss": 0.0681, "step": 15020 }, { - "epoch": 0.83, - "learning_rate": 3.333888611249931e-06, - "loss": 0.1155, + "epoch": 1.67, + "learning_rate": 3.3307418924922257e-06, + "loss": 0.0968, "step": 15025 }, { - "epoch": 0.83, - "learning_rate": 3.3283358320839583e-06, - "loss": 0.0455, + "epoch": 1.67, + "learning_rate": 3.3251888049755664e-06, + "loss": 0.0695, "step": 15030 }, { - "epoch": 0.83, - "learning_rate": 3.3227830529179856e-06, - "loss": 0.072, + "epoch": 1.67, + "learning_rate": 3.3196357174589074e-06, + "loss": 0.0358, "step": 15035 }, { - "epoch": 0.83, - "learning_rate": 3.317230273752013e-06, - "loss": 0.1365, + "epoch": 1.67, + "learning_rate": 3.314082629942248e-06, + "loss": 0.0515, "step": 15040 }, { - "epoch": 0.83, - "learning_rate": 3.311677494586041e-06, - "loss": 0.099, + "epoch": 1.67, + "learning_rate": 3.3085295424255887e-06, + "loss": 0.0757, "step": 15045 }, { - "epoch": 0.84, - "learning_rate": 3.306124715420068e-06, - "loss": 0.2399, + "epoch": 1.67, + "learning_rate": 3.3029764549089294e-06, + "loss": 0.1106, "step": 15050 }, { - "epoch": 0.84, - "learning_rate": 3.3005719362540955e-06, - "loss": 0.1116, + "epoch": 1.67, + "learning_rate": 3.297423367392271e-06, + "loss": 0.1858, "step": 15055 }, { - "epoch": 0.84, - "learning_rate": 3.295019157088123e-06, - "loss": 0.1252, + "epoch": 1.67, + "learning_rate": 3.2918702798756115e-06, + "loss": 0.2579, "step": 15060 }, { - "epoch": 0.84, - "learning_rate": 3.2894663779221503e-06, - "loss": 0.0956, + "epoch": 1.67, + "learning_rate": 3.286317192358952e-06, + "loss": 0.0849, "step": 15065 }, { - "epoch": 0.84, - "learning_rate": 3.2839135987561777e-06, - "loss": 0.138, + "epoch": 1.67, + "learning_rate": 3.2807641048422928e-06, + "loss": 0.0685, "step": 15070 }, { - "epoch": 0.84, - "learning_rate": 3.278360819590205e-06, - "loss": 0.0822, + "epoch": 1.67, + "learning_rate": 3.2752110173256334e-06, + "loss": 0.0719, "step": 15075 }, { - "epoch": 0.84, - "learning_rate": 3.2728080404242324e-06, - "loss": 0.0679, + "epoch": 1.67, + "learning_rate": 3.269657929808974e-06, + "loss": 0.0556, "step": 15080 }, { - "epoch": 0.84, - "learning_rate": 3.2672552612582598e-06, - "loss": 0.0745, + "epoch": 1.67, + "learning_rate": 3.264104842292315e-06, + "loss": 0.0503, "step": 15085 }, { - "epoch": 0.84, - "learning_rate": 3.2617024820922876e-06, - "loss": 0.1312, + "epoch": 1.67, + "learning_rate": 3.2585517547756557e-06, + "loss": 0.0477, "step": 15090 }, { - "epoch": 0.84, - "learning_rate": 3.256149702926315e-06, - "loss": 0.202, + "epoch": 1.68, + "learning_rate": 3.2529986672589964e-06, + "loss": 0.1316, "step": 15095 }, { - "epoch": 0.84, - "learning_rate": 3.2505969237603423e-06, - "loss": 0.3009, + "epoch": 1.68, + "learning_rate": 3.247445579742337e-06, + "loss": 0.0517, "step": 15100 }, { - "epoch": 0.84, - "learning_rate": 3.2450441445943697e-06, - "loss": 0.0967, + "epoch": 1.68, + "learning_rate": 3.2418924922256776e-06, + "loss": 0.1516, "step": 15105 }, { - "epoch": 0.84, - "learning_rate": 3.239491365428397e-06, - "loss": 0.0722, + "epoch": 1.68, + "learning_rate": 3.2363394047090187e-06, + "loss": 0.2206, "step": 15110 }, { - "epoch": 0.84, - "learning_rate": 3.2339385862624244e-06, - "loss": 0.0651, + "epoch": 1.68, + "learning_rate": 3.2307863171923594e-06, + "loss": 0.0522, "step": 15115 }, { - "epoch": 0.84, - "learning_rate": 3.228385807096452e-06, - "loss": 0.1045, + "epoch": 1.68, + "learning_rate": 3.2252332296757e-06, + "loss": 0.0559, "step": 15120 }, { - "epoch": 0.84, - "learning_rate": 3.222833027930479e-06, - "loss": 0.0861, + "epoch": 1.68, + "learning_rate": 3.2196801421590406e-06, + "loss": 0.0735, "step": 15125 }, { - "epoch": 0.84, - "learning_rate": 3.217280248764507e-06, - "loss": 0.052, + "epoch": 1.68, + "learning_rate": 3.2141270546423813e-06, + "loss": 0.0775, "step": 15130 }, { - "epoch": 0.84, - "learning_rate": 3.2117274695985343e-06, - "loss": 0.089, + "epoch": 1.68, + "learning_rate": 3.208573967125722e-06, + "loss": 0.0308, "step": 15135 }, { - "epoch": 0.84, - "learning_rate": 3.2061746904325617e-06, - "loss": 0.1757, + "epoch": 1.68, + "learning_rate": 3.203020879609063e-06, + "loss": 0.0385, "step": 15140 }, { - "epoch": 0.84, - "learning_rate": 3.200621911266589e-06, - "loss": 0.1147, + "epoch": 1.68, + "learning_rate": 3.1974677920924036e-06, + "loss": 0.0634, "step": 15145 }, { - "epoch": 0.84, - "learning_rate": 3.1950691321006165e-06, - "loss": 0.3078, + "epoch": 1.68, + "learning_rate": 3.1919147045757442e-06, + "loss": 0.0936, "step": 15150 }, { - "epoch": 0.84, - "learning_rate": 3.189516352934644e-06, - "loss": 0.098, + "epoch": 1.68, + "learning_rate": 3.186361617059085e-06, + "loss": 0.126, "step": 15155 }, { - "epoch": 0.84, - "learning_rate": 3.183963573768671e-06, - "loss": 0.0898, + "epoch": 1.68, + "learning_rate": 3.1808085295424255e-06, + "loss": 0.2512, "step": 15160 }, { - "epoch": 0.84, - "learning_rate": 3.1784107946026986e-06, - "loss": 0.1019, + "epoch": 1.68, + "learning_rate": 3.1752554420257666e-06, + "loss": 0.0554, "step": 15165 }, { - "epoch": 0.84, - "learning_rate": 3.172858015436726e-06, - "loss": 0.1207, + "epoch": 1.68, + "learning_rate": 3.1697023545091072e-06, + "loss": 0.062, "step": 15170 }, { - "epoch": 0.84, - "learning_rate": 3.167305236270754e-06, - "loss": 0.1103, + "epoch": 1.68, + "learning_rate": 3.164149266992448e-06, + "loss": 0.0806, "step": 15175 }, { - "epoch": 0.84, - "learning_rate": 3.1617524571047815e-06, - "loss": 0.0473, + "epoch": 1.68, + "learning_rate": 3.1585961794757885e-06, + "loss": 0.0779, "step": 15180 }, { - "epoch": 0.84, - "learning_rate": 3.156199677938809e-06, - "loss": 0.1305, + "epoch": 1.69, + "learning_rate": 3.153043091959129e-06, + "loss": 0.0697, "step": 15185 }, { - "epoch": 0.84, - "learning_rate": 3.1506468987728363e-06, - "loss": 0.1091, + "epoch": 1.69, + "learning_rate": 3.1474900044424706e-06, + "loss": 0.0491, "step": 15190 }, { - "epoch": 0.84, - "learning_rate": 3.1450941196068637e-06, - "loss": 0.1349, + "epoch": 1.69, + "learning_rate": 3.1419369169258113e-06, + "loss": 0.0758, "step": 15195 }, { - "epoch": 0.84, - "learning_rate": 3.139541340440891e-06, - "loss": 0.424, + "epoch": 1.69, + "learning_rate": 3.136383829409152e-06, + "loss": 0.0753, "step": 15200 }, { - "epoch": 0.84, - "learning_rate": 3.1339885612749184e-06, - "loss": 0.1027, + "epoch": 1.69, + "learning_rate": 3.1308307418924925e-06, + "loss": 0.1884, "step": 15205 }, { - "epoch": 0.84, - "learning_rate": 3.1284357821089458e-06, - "loss": 0.1191, + "epoch": 1.69, + "learning_rate": 3.1252776543758336e-06, + "loss": 0.1694, "step": 15210 }, { - "epoch": 0.84, - "learning_rate": 3.122883002942973e-06, - "loss": 0.1037, + "epoch": 1.69, + "learning_rate": 3.1197245668591742e-06, + "loss": 0.0792, "step": 15215 }, { - "epoch": 0.84, - "learning_rate": 3.117330223777001e-06, - "loss": 0.0899, + "epoch": 1.69, + "learning_rate": 3.114171479342515e-06, + "loss": 0.0557, "step": 15220 }, { - "epoch": 0.84, - "learning_rate": 3.1117774446110283e-06, - "loss": 0.0786, + "epoch": 1.69, + "learning_rate": 3.1086183918258555e-06, + "loss": 0.0857, "step": 15225 }, { - "epoch": 0.85, - "learning_rate": 3.1062246654450557e-06, - "loss": 0.0549, + "epoch": 1.69, + "learning_rate": 3.103065304309196e-06, + "loss": 0.083, "step": 15230 }, { - "epoch": 0.85, - "learning_rate": 3.100671886279083e-06, - "loss": 0.1529, + "epoch": 1.69, + "learning_rate": 3.097512216792537e-06, + "loss": 0.0444, "step": 15235 }, { - "epoch": 0.85, - "learning_rate": 3.0951191071131104e-06, - "loss": 0.065, + "epoch": 1.69, + "learning_rate": 3.091959129275878e-06, + "loss": 0.0365, "step": 15240 }, { - "epoch": 0.85, - "learning_rate": 3.089566327947138e-06, - "loss": 0.1782, + "epoch": 1.69, + "learning_rate": 3.0864060417592185e-06, + "loss": 0.0488, "step": 15245 }, { - "epoch": 0.85, - "learning_rate": 3.084013548781165e-06, - "loss": 0.3391, + "epoch": 1.69, + "learning_rate": 3.080852954242559e-06, + "loss": 0.0521, "step": 15250 }, { - "epoch": 0.85, - "learning_rate": 3.0784607696151925e-06, - "loss": 0.0982, + "epoch": 1.69, + "learning_rate": 3.0752998667258998e-06, + "loss": 0.1197, "step": 15255 }, { - "epoch": 0.85, - "learning_rate": 3.07290799044922e-06, - "loss": 0.0601, + "epoch": 1.69, + "learning_rate": 3.0697467792092404e-06, + "loss": 0.2226, "step": 15260 }, { - "epoch": 0.85, - "learning_rate": 3.0673552112832477e-06, - "loss": 0.1164, + "epoch": 1.69, + "learning_rate": 3.0641936916925815e-06, + "loss": 0.0647, "step": 15265 }, { - "epoch": 0.85, - "learning_rate": 3.061802432117275e-06, - "loss": 0.121, + "epoch": 1.69, + "learning_rate": 3.058640604175922e-06, + "loss": 0.0681, "step": 15270 }, { - "epoch": 0.85, - "learning_rate": 3.0562496529513025e-06, - "loss": 0.0731, + "epoch": 1.7, + "learning_rate": 3.0530875166592628e-06, + "loss": 0.0726, "step": 15275 }, { - "epoch": 0.85, - "learning_rate": 3.05069687378533e-06, - "loss": 0.0391, + "epoch": 1.7, + "learning_rate": 3.0475344291426034e-06, + "loss": 0.054, "step": 15280 }, { - "epoch": 0.85, - "learning_rate": 3.045144094619357e-06, - "loss": 0.0991, + "epoch": 1.7, + "learning_rate": 3.041981341625944e-06, + "loss": 0.0524, "step": 15285 }, { - "epoch": 0.85, - "learning_rate": 3.0395913154533846e-06, - "loss": 0.0988, + "epoch": 1.7, + "learning_rate": 3.0364282541092847e-06, + "loss": 0.0373, "step": 15290 }, { - "epoch": 0.85, - "learning_rate": 3.0351490921206066e-06, - "loss": 1.0191, + "epoch": 1.7, + "learning_rate": 3.0308751665926257e-06, + "loss": 0.0549, "step": 15295 }, { - "epoch": 0.85, - "learning_rate": 3.029596312954634e-06, - "loss": 0.2898, + "epoch": 1.7, + "learning_rate": 3.0253220790759664e-06, + "loss": 0.0793, "step": 15300 }, { - "epoch": 0.85, - "learning_rate": 3.0240435337886614e-06, - "loss": 0.0959, + "epoch": 1.7, + "learning_rate": 3.019768991559307e-06, + "loss": 0.1343, "step": 15305 }, { - "epoch": 0.85, - "learning_rate": 3.0184907546226888e-06, - "loss": 0.0998, + "epoch": 1.7, + "learning_rate": 3.0142159040426477e-06, + "loss": 0.293, "step": 15310 }, { - "epoch": 0.85, - "learning_rate": 3.012937975456716e-06, - "loss": 0.1218, + "epoch": 1.7, + "learning_rate": 3.0086628165259883e-06, + "loss": 0.0535, "step": 15315 }, { - "epoch": 0.85, - "learning_rate": 3.0073851962907435e-06, - "loss": 0.12, + "epoch": 1.7, + "learning_rate": 3.0031097290093294e-06, + "loss": 0.0655, "step": 15320 }, { - "epoch": 0.85, - "learning_rate": 3.0018324171247713e-06, - "loss": 0.0814, + "epoch": 1.7, + "learning_rate": 2.99755664149267e-06, + "loss": 0.0613, "step": 15325 }, { - "epoch": 0.85, - "learning_rate": 2.9962796379587987e-06, - "loss": 0.0872, + "epoch": 1.7, + "learning_rate": 2.992003553976011e-06, + "loss": 0.079, "step": 15330 }, { - "epoch": 0.85, - "learning_rate": 2.990726858792826e-06, - "loss": 0.0896, + "epoch": 1.7, + "learning_rate": 2.9864504664593517e-06, + "loss": 0.0452, "step": 15335 }, { - "epoch": 0.85, - "learning_rate": 2.9851740796268534e-06, - "loss": 0.1132, + "epoch": 1.7, + "learning_rate": 2.9808973789426928e-06, + "loss": 0.0312, "step": 15340 }, { - "epoch": 0.85, - "learning_rate": 2.979621300460881e-06, - "loss": 0.1891, + "epoch": 1.7, + "learning_rate": 2.9753442914260334e-06, + "loss": 0.0569, "step": 15345 }, { - "epoch": 0.85, - "learning_rate": 2.974068521294908e-06, - "loss": 0.299, + "epoch": 1.7, + "learning_rate": 2.969791203909374e-06, + "loss": 0.0673, "step": 15350 }, { - "epoch": 0.85, - "learning_rate": 2.9685157421289355e-06, - "loss": 0.0913, + "epoch": 1.7, + "learning_rate": 2.9642381163927147e-06, + "loss": 0.1319, "step": 15355 }, { - "epoch": 0.85, - "learning_rate": 2.962962962962963e-06, - "loss": 0.0827, + "epoch": 1.7, + "learning_rate": 2.9586850288760553e-06, + "loss": 0.2331, "step": 15360 }, { - "epoch": 0.85, - "learning_rate": 2.9574101837969903e-06, - "loss": 0.1346, + "epoch": 1.71, + "learning_rate": 2.9531319413593964e-06, + "loss": 0.0618, "step": 15365 }, { - "epoch": 0.85, - "learning_rate": 2.9518574046310185e-06, - "loss": 0.122, + "epoch": 1.71, + "learning_rate": 2.947578853842737e-06, + "loss": 0.0528, "step": 15370 }, { - "epoch": 0.85, - "learning_rate": 2.946304625465046e-06, - "loss": 0.0831, + "epoch": 1.71, + "learning_rate": 2.9420257663260776e-06, + "loss": 0.0505, "step": 15375 }, { - "epoch": 0.85, - "learning_rate": 2.9407518462990732e-06, - "loss": 0.0567, + "epoch": 1.71, + "learning_rate": 2.9364726788094183e-06, + "loss": 0.0808, "step": 15380 }, { - "epoch": 0.85, - "learning_rate": 2.9351990671331006e-06, - "loss": 0.1747, + "epoch": 1.71, + "learning_rate": 2.930919591292759e-06, + "loss": 0.0549, "step": 15385 }, { - "epoch": 0.85, - "learning_rate": 2.929646287967128e-06, - "loss": 0.0893, + "epoch": 1.71, + "learning_rate": 2.9253665037760996e-06, + "loss": 0.022, "step": 15390 }, { - "epoch": 0.85, - "learning_rate": 2.9240935088011554e-06, - "loss": 0.2117, + "epoch": 1.71, + "learning_rate": 2.9198134162594406e-06, + "loss": 0.0489, "step": 15395 }, { - "epoch": 0.85, - "learning_rate": 2.9185407296351827e-06, - "loss": 0.3089, + "epoch": 1.71, + "learning_rate": 2.9142603287427813e-06, + "loss": 0.103, "step": 15400 }, { - "epoch": 0.85, - "learning_rate": 2.91298795046921e-06, - "loss": 0.0822, + "epoch": 1.71, + "learning_rate": 2.908707241226122e-06, + "loss": 0.1584, "step": 15405 }, { - "epoch": 0.86, - "learning_rate": 2.9074351713032375e-06, - "loss": 0.1538, + "epoch": 1.71, + "learning_rate": 2.9031541537094625e-06, + "loss": 0.2287, "step": 15410 }, { - "epoch": 0.86, - "learning_rate": 2.9018823921372653e-06, - "loss": 0.1083, + "epoch": 1.71, + "learning_rate": 2.897601066192803e-06, + "loss": 0.0567, "step": 15415 }, { - "epoch": 0.86, - "learning_rate": 2.8963296129712926e-06, - "loss": 0.0788, + "epoch": 1.71, + "learning_rate": 2.8920479786761442e-06, + "loss": 0.0645, "step": 15420 }, { - "epoch": 0.86, - "learning_rate": 2.89077683380532e-06, - "loss": 0.1918, + "epoch": 1.71, + "learning_rate": 2.886494891159485e-06, + "loss": 0.0543, "step": 15425 }, { - "epoch": 0.86, - "learning_rate": 2.8852240546393474e-06, - "loss": 0.0559, + "epoch": 1.71, + "learning_rate": 2.8809418036428255e-06, + "loss": 0.1019, "step": 15430 }, { - "epoch": 0.86, - "learning_rate": 2.8796712754733748e-06, - "loss": 0.0977, + "epoch": 1.71, + "learning_rate": 2.875388716126166e-06, + "loss": 0.0368, "step": 15435 }, { - "epoch": 0.86, - "learning_rate": 2.874118496307402e-06, - "loss": 0.1205, + "epoch": 1.71, + "learning_rate": 2.869835628609507e-06, + "loss": 0.0472, "step": 15440 }, { - "epoch": 0.86, - "learning_rate": 2.8685657171414295e-06, - "loss": 0.1882, + "epoch": 1.71, + "learning_rate": 2.8642825410928474e-06, + "loss": 0.1049, "step": 15445 }, { - "epoch": 0.86, - "learning_rate": 2.863012937975457e-06, - "loss": 0.2089, + "epoch": 1.71, + "learning_rate": 2.8587294535761885e-06, + "loss": 0.0807, "step": 15450 }, { - "epoch": 0.86, - "learning_rate": 2.8574601588094843e-06, - "loss": 0.0904, + "epoch": 1.72, + "learning_rate": 2.853176366059529e-06, + "loss": 0.1651, "step": 15455 }, { - "epoch": 0.86, - "learning_rate": 2.851907379643512e-06, - "loss": 0.0945, + "epoch": 1.72, + "learning_rate": 2.8476232785428698e-06, + "loss": 0.2318, "step": 15460 }, { - "epoch": 0.86, - "learning_rate": 2.8463546004775394e-06, - "loss": 0.1342, + "epoch": 1.72, + "learning_rate": 2.8420701910262113e-06, + "loss": 0.0673, "step": 15465 }, { - "epoch": 0.86, - "learning_rate": 2.840801821311567e-06, - "loss": 0.0752, + "epoch": 1.72, + "learning_rate": 2.836517103509552e-06, + "loss": 0.0686, "step": 15470 }, { - "epoch": 0.86, - "learning_rate": 2.835249042145594e-06, - "loss": 0.1056, + "epoch": 1.72, + "learning_rate": 2.8309640159928925e-06, + "loss": 0.0521, "step": 15475 }, { - "epoch": 0.86, - "learning_rate": 2.8296962629796215e-06, - "loss": 0.0725, + "epoch": 1.72, + "learning_rate": 2.825410928476233e-06, + "loss": 0.0924, "step": 15480 }, { - "epoch": 0.86, - "learning_rate": 2.824143483813649e-06, - "loss": 0.0852, + "epoch": 1.72, + "learning_rate": 2.819857840959574e-06, + "loss": 0.0463, "step": 15485 }, { - "epoch": 0.86, - "learning_rate": 2.8185907046476763e-06, - "loss": 0.1252, + "epoch": 1.72, + "learning_rate": 2.8143047534429145e-06, + "loss": 0.0263, "step": 15490 }, { - "epoch": 0.86, - "learning_rate": 2.8130379254817037e-06, - "loss": 0.1616, + "epoch": 1.72, + "learning_rate": 2.8087516659262555e-06, + "loss": 0.0828, "step": 15495 }, { - "epoch": 0.86, - "learning_rate": 2.807485146315731e-06, - "loss": 0.5112, + "epoch": 1.72, + "learning_rate": 2.803198578409596e-06, + "loss": 0.0665, "step": 15500 }, { - "epoch": 0.86, - "learning_rate": 2.801932367149759e-06, - "loss": 0.1017, + "epoch": 1.72, + "learning_rate": 2.797645490892937e-06, + "loss": 0.153, "step": 15505 }, { - "epoch": 0.86, - "learning_rate": 2.796379587983786e-06, - "loss": 0.096, + "epoch": 1.72, + "learning_rate": 2.7920924033762774e-06, + "loss": 0.1715, "step": 15510 }, { - "epoch": 0.86, - "learning_rate": 2.7908268088178136e-06, - "loss": 0.1028, + "epoch": 1.72, + "learning_rate": 2.786539315859618e-06, + "loss": 0.0524, "step": 15515 }, { - "epoch": 0.86, - "learning_rate": 2.785274029651841e-06, - "loss": 0.131, + "epoch": 1.72, + "learning_rate": 2.780986228342959e-06, + "loss": 0.0592, "step": 15520 }, { - "epoch": 0.86, - "learning_rate": 2.7797212504858683e-06, - "loss": 0.0658, + "epoch": 1.72, + "learning_rate": 2.7754331408262998e-06, + "loss": 0.0889, "step": 15525 }, { - "epoch": 0.86, - "learning_rate": 2.7741684713198957e-06, - "loss": 0.0864, + "epoch": 1.72, + "learning_rate": 2.7698800533096404e-06, + "loss": 0.076, "step": 15530 }, { - "epoch": 0.86, - "learning_rate": 2.768615692153923e-06, - "loss": 0.1366, + "epoch": 1.72, + "learning_rate": 2.764326965792981e-06, + "loss": 0.062, "step": 15535 }, { - "epoch": 0.86, - "learning_rate": 2.7630629129879504e-06, - "loss": 0.0553, + "epoch": 1.72, + "learning_rate": 2.7587738782763217e-06, + "loss": 0.0371, "step": 15540 }, { - "epoch": 0.86, - "learning_rate": 2.757510133821978e-06, - "loss": 0.171, + "epoch": 1.73, + "learning_rate": 2.7532207907596623e-06, + "loss": 0.074, "step": 15545 }, { - "epoch": 0.86, - "learning_rate": 2.7519573546560056e-06, - "loss": 0.2554, + "epoch": 1.73, + "learning_rate": 2.7476677032430034e-06, + "loss": 0.0894, "step": 15550 }, { - "epoch": 0.86, - "learning_rate": 2.746404575490033e-06, - "loss": 0.0971, + "epoch": 1.73, + "learning_rate": 2.742114615726344e-06, + "loss": 0.1149, "step": 15555 }, { - "epoch": 0.86, - "learning_rate": 2.7408517963240603e-06, - "loss": 0.0875, + "epoch": 1.73, + "learning_rate": 2.7365615282096847e-06, + "loss": 0.2586, "step": 15560 }, { - "epoch": 0.86, - "learning_rate": 2.7352990171580877e-06, - "loss": 0.1034, + "epoch": 1.73, + "learning_rate": 2.7310084406930253e-06, + "loss": 0.0424, "step": 15565 }, { - "epoch": 0.86, - "learning_rate": 2.729746237992115e-06, - "loss": 0.1486, + "epoch": 1.73, + "learning_rate": 2.725455353176366e-06, + "loss": 0.0762, "step": 15570 }, { - "epoch": 0.86, - "learning_rate": 2.7241934588261425e-06, - "loss": 0.0701, + "epoch": 1.73, + "learning_rate": 2.719902265659707e-06, + "loss": 0.065, "step": 15575 }, { - "epoch": 0.86, - "learning_rate": 2.71864067966017e-06, - "loss": 0.0616, + "epoch": 1.73, + "learning_rate": 2.7143491781430477e-06, + "loss": 0.065, "step": 15580 }, { - "epoch": 0.86, - "learning_rate": 2.713087900494197e-06, - "loss": 0.0952, + "epoch": 1.73, + "learning_rate": 2.7087960906263883e-06, + "loss": 0.0514, "step": 15585 }, { - "epoch": 0.87, - "learning_rate": 2.7075351213282254e-06, - "loss": 0.0793, + "epoch": 1.73, + "learning_rate": 2.703243003109729e-06, + "loss": 0.0525, "step": 15590 }, { - "epoch": 0.87, - "learning_rate": 2.701982342162253e-06, - "loss": 0.0988, + "epoch": 1.73, + "learning_rate": 2.6976899155930696e-06, + "loss": 0.0781, "step": 15595 }, { - "epoch": 0.87, - "learning_rate": 2.69642956299628e-06, - "loss": 0.4452, + "epoch": 1.73, + "learning_rate": 2.692136828076411e-06, + "loss": 0.065, "step": 15600 }, { - "epoch": 0.87, - "learning_rate": 2.6908767838303075e-06, - "loss": 0.0862, + "epoch": 1.73, + "learning_rate": 2.6865837405597517e-06, + "loss": 0.1071, "step": 15605 }, { - "epoch": 0.87, - "learning_rate": 2.685324004664335e-06, - "loss": 0.1002, + "epoch": 1.73, + "learning_rate": 2.6810306530430923e-06, + "loss": 0.264, "step": 15610 }, { - "epoch": 0.87, - "learning_rate": 2.6797712254983623e-06, - "loss": 0.0697, + "epoch": 1.73, + "learning_rate": 2.675477565526433e-06, + "loss": 0.0622, "step": 15615 }, { - "epoch": 0.87, - "learning_rate": 2.6742184463323897e-06, - "loss": 0.1231, + "epoch": 1.73, + "learning_rate": 2.669924478009774e-06, + "loss": 0.069, "step": 15620 }, { - "epoch": 0.87, - "learning_rate": 2.668665667166417e-06, - "loss": 0.0707, + "epoch": 1.73, + "learning_rate": 2.6643713904931147e-06, + "loss": 0.0867, "step": 15625 }, { - "epoch": 0.87, - "learning_rate": 2.6631128880004444e-06, - "loss": 0.0711, + "epoch": 1.73, + "learning_rate": 2.6588183029764553e-06, + "loss": 0.0804, "step": 15630 }, { - "epoch": 0.87, - "learning_rate": 2.657560108834472e-06, - "loss": 0.0848, + "epoch": 1.74, + "learning_rate": 2.653265215459796e-06, + "loss": 0.0436, "step": 15635 }, { - "epoch": 0.87, - "learning_rate": 2.6520073296684996e-06, - "loss": 0.109, + "epoch": 1.74, + "learning_rate": 2.6477121279431366e-06, + "loss": 0.043, "step": 15640 }, { - "epoch": 0.87, - "learning_rate": 2.646454550502527e-06, - "loss": 0.1598, + "epoch": 1.74, + "learning_rate": 2.6421590404264772e-06, + "loss": 0.0712, "step": 15645 }, { - "epoch": 0.87, - "learning_rate": 2.6409017713365543e-06, - "loss": 0.232, + "epoch": 1.74, + "learning_rate": 2.6366059529098183e-06, + "loss": 0.1312, "step": 15650 }, { - "epoch": 0.87, - "learning_rate": 2.6353489921705817e-06, - "loss": 0.1115, + "epoch": 1.74, + "learning_rate": 2.631052865393159e-06, + "loss": 0.1397, "step": 15655 }, { - "epoch": 0.87, - "learning_rate": 2.629796213004609e-06, - "loss": 0.0559, + "epoch": 1.74, + "learning_rate": 2.6254997778764996e-06, + "loss": 0.4137, "step": 15660 }, { - "epoch": 0.87, - "learning_rate": 2.6242434338386364e-06, - "loss": 0.1088, + "epoch": 1.74, + "learning_rate": 2.61994669035984e-06, + "loss": 0.0676, "step": 15665 }, { - "epoch": 0.87, - "learning_rate": 2.618690654672664e-06, - "loss": 0.1096, + "epoch": 1.74, + "learning_rate": 2.614393602843181e-06, + "loss": 0.0832, "step": 15670 }, { - "epoch": 0.87, - "learning_rate": 2.613137875506691e-06, - "loss": 0.0655, + "epoch": 1.74, + "learning_rate": 2.608840515326522e-06, + "loss": 0.077, "step": 15675 }, { - "epoch": 0.87, - "learning_rate": 2.607585096340719e-06, - "loss": 0.014, + "epoch": 1.74, + "learning_rate": 2.6032874278098625e-06, + "loss": 0.075, "step": 15680 }, { - "epoch": 0.87, - "learning_rate": 2.6020323171747463e-06, - "loss": 0.0645, + "epoch": 1.74, + "learning_rate": 2.597734340293203e-06, + "loss": 0.0253, "step": 15685 }, { - "epoch": 0.87, - "learning_rate": 2.5964795380087737e-06, - "loss": 0.1102, + "epoch": 1.74, + "learning_rate": 2.592181252776544e-06, + "loss": 0.0755, "step": 15690 }, { - "epoch": 0.87, - "learning_rate": 2.590926758842801e-06, - "loss": 0.1852, + "epoch": 1.74, + "learning_rate": 2.5866281652598845e-06, + "loss": 0.1044, "step": 15695 }, { - "epoch": 0.87, - "learning_rate": 2.5853739796768285e-06, - "loss": 0.323, + "epoch": 1.74, + "learning_rate": 2.581075077743225e-06, + "loss": 0.1227, "step": 15700 }, { - "epoch": 0.87, - "learning_rate": 2.579821200510856e-06, - "loss": 0.0837, + "epoch": 1.74, + "learning_rate": 2.575521990226566e-06, + "loss": 0.212, "step": 15705 }, { - "epoch": 0.87, - "learning_rate": 2.574268421344883e-06, - "loss": 0.1182, + "epoch": 1.74, + "learning_rate": 2.569968902709907e-06, + "loss": 0.2274, "step": 15710 }, { - "epoch": 0.87, - "learning_rate": 2.5687156421789106e-06, - "loss": 0.1172, + "epoch": 1.74, + "learning_rate": 2.5644158151932474e-06, + "loss": 0.0595, "step": 15715 }, { - "epoch": 0.87, - "learning_rate": 2.563162863012938e-06, - "loss": 0.1243, + "epoch": 1.74, + "learning_rate": 2.558862727676588e-06, + "loss": 0.0656, "step": 15720 }, { - "epoch": 0.87, - "learning_rate": 2.5576100838469657e-06, - "loss": 0.1612, + "epoch": 1.75, + "learning_rate": 2.5533096401599287e-06, + "loss": 0.0786, "step": 15725 }, { - "epoch": 0.87, - "learning_rate": 2.552057304680993e-06, - "loss": 0.0396, + "epoch": 1.75, + "learning_rate": 2.5477565526432698e-06, + "loss": 0.1025, "step": 15730 }, { - "epoch": 0.87, - "learning_rate": 2.5465045255150205e-06, - "loss": 0.0809, + "epoch": 1.75, + "learning_rate": 2.542203465126611e-06, + "loss": 0.0454, "step": 15735 }, { - "epoch": 0.87, - "learning_rate": 2.540951746349048e-06, - "loss": 0.1443, + "epoch": 1.75, + "learning_rate": 2.5366503776099515e-06, + "loss": 0.0552, "step": 15740 }, { - "epoch": 0.87, - "learning_rate": 2.5353989671830752e-06, - "loss": 0.1872, + "epoch": 1.75, + "learning_rate": 2.531097290093292e-06, + "loss": 0.087, "step": 15745 }, { - "epoch": 0.87, - "learning_rate": 2.5298461880171026e-06, - "loss": 0.2681, + "epoch": 1.75, + "learning_rate": 2.525544202576633e-06, + "loss": 0.1139, "step": 15750 }, { - "epoch": 0.87, - "learning_rate": 2.52429340885113e-06, - "loss": 0.0661, + "epoch": 1.75, + "learning_rate": 2.519991115059974e-06, + "loss": 0.1452, "step": 15755 }, { - "epoch": 0.87, - "learning_rate": 2.5187406296851573e-06, - "loss": 0.0867, + "epoch": 1.75, + "learning_rate": 2.5144380275433145e-06, + "loss": 0.2648, "step": 15760 }, { - "epoch": 0.87, - "learning_rate": 2.5131878505191847e-06, - "loss": 0.0963, + "epoch": 1.75, + "learning_rate": 2.508884940026655e-06, + "loss": 0.0462, "step": 15765 }, { - "epoch": 0.88, - "learning_rate": 2.5076350713532125e-06, - "loss": 0.0862, + "epoch": 1.75, + "learning_rate": 2.5033318525099957e-06, + "loss": 0.0516, "step": 15770 }, { - "epoch": 0.88, - "learning_rate": 2.50208229218724e-06, - "loss": 0.0841, + "epoch": 1.75, + "learning_rate": 2.497778764993337e-06, + "loss": 0.0783, "step": 15775 }, { - "epoch": 0.88, - "learning_rate": 2.4965295130212673e-06, - "loss": 0.0289, + "epoch": 1.75, + "learning_rate": 2.4922256774766774e-06, + "loss": 0.0758, "step": 15780 }, { - "epoch": 0.88, - "learning_rate": 2.4909767338552946e-06, - "loss": 0.0692, + "epoch": 1.75, + "learning_rate": 2.486672589960018e-06, + "loss": 0.0221, "step": 15785 }, { - "epoch": 0.88, - "learning_rate": 2.485423954689322e-06, - "loss": 0.0972, + "epoch": 1.75, + "learning_rate": 2.4811195024433587e-06, + "loss": 0.0403, "step": 15790 }, { - "epoch": 0.88, - "learning_rate": 2.4798711755233494e-06, - "loss": 0.1081, + "epoch": 1.75, + "learning_rate": 2.4755664149266994e-06, + "loss": 0.0741, "step": 15795 }, { - "epoch": 0.88, - "learning_rate": 2.474318396357377e-06, - "loss": 0.2879, + "epoch": 1.75, + "learning_rate": 2.47001332741004e-06, + "loss": 0.1426, "step": 15800 }, { - "epoch": 0.88, - "learning_rate": 2.4687656171914045e-06, - "loss": 0.1107, + "epoch": 1.75, + "learning_rate": 2.464460239893381e-06, + "loss": 0.1684, "step": 15805 }, { - "epoch": 0.88, - "learning_rate": 2.463212838025432e-06, - "loss": 0.0889, + "epoch": 1.75, + "learning_rate": 2.4589071523767217e-06, + "loss": 0.2842, "step": 15810 }, { - "epoch": 0.88, - "learning_rate": 2.4576600588594593e-06, - "loss": 0.09, + "epoch": 1.76, + "learning_rate": 2.4533540648600623e-06, + "loss": 0.064, "step": 15815 }, { - "epoch": 0.88, - "learning_rate": 2.4521072796934867e-06, - "loss": 0.1092, + "epoch": 1.76, + "learning_rate": 2.447800977343403e-06, + "loss": 0.0618, "step": 15820 }, { - "epoch": 0.88, - "learning_rate": 2.4465545005275145e-06, - "loss": 0.1299, + "epoch": 1.76, + "learning_rate": 2.4422478898267436e-06, + "loss": 0.0714, "step": 15825 }, { - "epoch": 0.88, - "learning_rate": 2.441001721361542e-06, - "loss": 0.1196, + "epoch": 1.76, + "learning_rate": 2.4366948023100847e-06, + "loss": 0.082, "step": 15830 }, { - "epoch": 0.88, - "learning_rate": 2.435448942195569e-06, - "loss": 0.0846, + "epoch": 1.76, + "learning_rate": 2.4311417147934253e-06, + "loss": 0.0634, "step": 15835 }, { - "epoch": 0.88, - "learning_rate": 2.4298961630295966e-06, - "loss": 0.1314, + "epoch": 1.76, + "learning_rate": 2.4255886272767664e-06, + "loss": 0.052, "step": 15840 }, { - "epoch": 0.88, - "learning_rate": 2.424343383863624e-06, - "loss": 0.1108, + "epoch": 1.76, + "learning_rate": 2.420035539760107e-06, + "loss": 0.05, "step": 15845 }, { - "epoch": 0.88, - "learning_rate": 2.4187906046976513e-06, - "loss": 0.4144, + "epoch": 1.76, + "learning_rate": 2.4144824522434477e-06, + "loss": 0.0859, "step": 15850 }, { - "epoch": 0.88, - "learning_rate": 2.4132378255316787e-06, - "loss": 0.0939, + "epoch": 1.76, + "learning_rate": 2.4089293647267883e-06, + "loss": 0.184, "step": 15855 }, { - "epoch": 0.88, - "learning_rate": 2.407685046365706e-06, - "loss": 0.0758, + "epoch": 1.76, + "learning_rate": 2.403376277210129e-06, + "loss": 0.196, "step": 15860 }, { - "epoch": 0.88, - "learning_rate": 2.4021322671997334e-06, - "loss": 0.1172, + "epoch": 1.76, + "learning_rate": 2.3978231896934696e-06, + "loss": 0.0579, "step": 15865 }, { - "epoch": 0.88, - "learning_rate": 2.3965794880337612e-06, - "loss": 0.1167, + "epoch": 1.76, + "learning_rate": 2.3922701021768106e-06, + "loss": 0.0851, "step": 15870 }, { - "epoch": 0.88, - "learning_rate": 2.3910267088677886e-06, - "loss": 0.1272, + "epoch": 1.76, + "learning_rate": 2.3867170146601513e-06, + "loss": 0.0778, "step": 15875 }, { - "epoch": 0.88, - "learning_rate": 2.385473929701816e-06, - "loss": 0.0757, + "epoch": 1.76, + "learning_rate": 2.381163927143492e-06, + "loss": 0.0696, "step": 15880 }, { - "epoch": 0.88, - "learning_rate": 2.3799211505358433e-06, - "loss": 0.0638, + "epoch": 1.76, + "learning_rate": 2.3756108396268325e-06, + "loss": 0.0442, "step": 15885 }, { - "epoch": 0.88, - "learning_rate": 2.3743683713698707e-06, - "loss": 0.1544, + "epoch": 1.76, + "learning_rate": 2.370057752110173e-06, + "loss": 0.0551, "step": 15890 }, { - "epoch": 0.88, - "learning_rate": 2.368815592203898e-06, - "loss": 0.17, + "epoch": 1.76, + "learning_rate": 2.3645046645935142e-06, + "loss": 0.0534, "step": 15895 }, { - "epoch": 0.88, - "learning_rate": 2.3632628130379255e-06, - "loss": 0.3077, + "epoch": 1.76, + "learning_rate": 2.358951577076855e-06, + "loss": 0.0989, "step": 15900 }, { - "epoch": 0.88, - "learning_rate": 2.357710033871953e-06, - "loss": 0.0771, + "epoch": 1.77, + "learning_rate": 2.353398489560196e-06, + "loss": 0.1461, "step": 15905 }, { - "epoch": 0.88, - "learning_rate": 2.3521572547059806e-06, - "loss": 0.0593, + "epoch": 1.77, + "learning_rate": 2.3478454020435366e-06, + "loss": 0.2242, "step": 15910 }, { - "epoch": 0.88, - "learning_rate": 2.346604475540008e-06, - "loss": 0.1094, + "epoch": 1.77, + "learning_rate": 2.3422923145268772e-06, + "loss": 0.0519, "step": 15915 }, { - "epoch": 0.88, - "learning_rate": 2.3410516963740354e-06, - "loss": 0.0944, + "epoch": 1.77, + "learning_rate": 2.336739227010218e-06, + "loss": 0.0797, "step": 15920 }, { - "epoch": 0.88, - "learning_rate": 2.3354989172080627e-06, - "loss": 0.1115, + "epoch": 1.77, + "learning_rate": 2.3311861394935585e-06, + "loss": 0.0454, "step": 15925 }, { - "epoch": 0.88, - "learning_rate": 2.32994613804209e-06, - "loss": 0.071, + "epoch": 1.77, + "learning_rate": 2.3256330519768996e-06, + "loss": 0.0969, "step": 15930 }, { - "epoch": 0.88, - "learning_rate": 2.324393358876118e-06, - "loss": 0.0622, + "epoch": 1.77, + "learning_rate": 2.32007996446024e-06, + "loss": 0.0963, "step": 15935 }, { - "epoch": 0.88, - "learning_rate": 2.3188405797101453e-06, - "loss": 0.0656, + "epoch": 1.77, + "learning_rate": 2.314526876943581e-06, + "loss": 0.0421, "step": 15940 }, { - "epoch": 0.88, - "learning_rate": 2.3132878005441727e-06, - "loss": 0.1092, + "epoch": 1.77, + "learning_rate": 2.3089737894269215e-06, + "loss": 0.0589, "step": 15945 }, { - "epoch": 0.89, - "learning_rate": 2.3077350213782e-06, - "loss": 0.2441, + "epoch": 1.77, + "learning_rate": 2.303420701910262e-06, + "loss": 0.0869, "step": 15950 }, { - "epoch": 0.89, - "learning_rate": 2.3021822422122274e-06, - "loss": 0.0704, + "epoch": 1.77, + "learning_rate": 2.2978676143936028e-06, + "loss": 0.1253, "step": 15955 }, { - "epoch": 0.89, - "learning_rate": 2.2966294630462548e-06, - "loss": 0.1161, + "epoch": 1.77, + "learning_rate": 2.292314526876944e-06, + "loss": 0.19, "step": 15960 }, { - "epoch": 0.89, - "learning_rate": 2.291076683880282e-06, - "loss": 0.1003, + "epoch": 1.77, + "learning_rate": 2.2867614393602845e-06, + "loss": 0.0704, "step": 15965 }, { - "epoch": 0.89, - "learning_rate": 2.2855239047143095e-06, - "loss": 0.0971, + "epoch": 1.77, + "learning_rate": 2.2812083518436255e-06, + "loss": 0.0893, "step": 15970 }, { - "epoch": 0.89, - "learning_rate": 2.279971125548337e-06, - "loss": 0.0757, + "epoch": 1.77, + "learning_rate": 2.275655264326966e-06, + "loss": 0.0607, "step": 15975 }, { - "epoch": 0.89, - "learning_rate": 2.2744183463823647e-06, - "loss": 0.0613, + "epoch": 1.77, + "learning_rate": 2.270102176810307e-06, + "loss": 0.0677, "step": 15980 }, { - "epoch": 0.89, - "learning_rate": 2.268865567216392e-06, - "loss": 0.0977, + "epoch": 1.77, + "learning_rate": 2.2645490892936474e-06, + "loss": 0.0351, "step": 15985 }, { - "epoch": 0.89, - "learning_rate": 2.2633127880504194e-06, - "loss": 0.1356, + "epoch": 1.77, + "learning_rate": 2.258996001776988e-06, + "loss": 0.0218, "step": 15990 }, { - "epoch": 0.89, - "learning_rate": 2.257760008884447e-06, - "loss": 0.1247, + "epoch": 1.78, + "learning_rate": 2.253442914260329e-06, + "loss": 0.0792, "step": 15995 }, { - "epoch": 0.89, - "learning_rate": 2.252207229718474e-06, - "loss": 0.3228, + "epoch": 1.78, + "learning_rate": 2.2478898267436698e-06, + "loss": 0.1009, "step": 16000 }, { - "epoch": 0.89, - "eval_loss": 0.04070247709751129, - "eval_runtime": 1393.3061, - "eval_samples_per_second": 11.496, - "eval_steps_per_second": 2.874, - "eval_wer": 0.10827623512441399, + "epoch": 1.78, + "eval_loss": 0.031791314482688904, + "eval_runtime": 1664.0089, + "eval_samples_per_second": 9.626, + "eval_steps_per_second": 2.407, + "eval_wer": 0.07012261089073206, "step": 16000 }, { - "epoch": 0.89, - "learning_rate": 2.2466544505525016e-06, - "loss": 0.1174, + "epoch": 1.78, + "learning_rate": 2.2423367392270104e-06, + "loss": 0.0741, "step": 16005 }, { - "epoch": 0.89, - "learning_rate": 2.241101671386529e-06, - "loss": 0.0835, + "epoch": 1.78, + "learning_rate": 2.236783651710351e-06, + "loss": 0.2546, "step": 16010 }, { - "epoch": 0.89, - "learning_rate": 2.2355488922205563e-06, - "loss": 0.0957, + "epoch": 1.78, + "learning_rate": 2.2312305641936917e-06, + "loss": 0.0593, "step": 16015 }, { - "epoch": 0.89, - "learning_rate": 2.229996113054584e-06, - "loss": 0.1212, + "epoch": 1.78, + "learning_rate": 2.2256774766770323e-06, + "loss": 0.0445, "step": 16020 }, { - "epoch": 0.89, - "learning_rate": 2.2244433338886115e-06, - "loss": 0.1128, + "epoch": 1.78, + "learning_rate": 2.2201243891603734e-06, + "loss": 0.0598, "step": 16025 }, { - "epoch": 0.89, - "learning_rate": 2.218890554722639e-06, - "loss": 0.0571, + "epoch": 1.78, + "learning_rate": 2.214571301643714e-06, + "loss": 0.0584, "step": 16030 }, { - "epoch": 0.89, - "learning_rate": 2.213337775556666e-06, - "loss": 0.0656, + "epoch": 1.78, + "learning_rate": 2.2090182141270547e-06, + "loss": 0.066, "step": 16035 }, { - "epoch": 0.89, - "learning_rate": 2.2077849963906936e-06, - "loss": 0.0609, + "epoch": 1.78, + "learning_rate": 2.2034651266103957e-06, + "loss": 0.0527, "step": 16040 }, { - "epoch": 0.89, - "learning_rate": 2.2022322172247214e-06, - "loss": 0.1588, + "epoch": 1.78, + "learning_rate": 2.1979120390937364e-06, + "loss": 0.0442, "step": 16045 }, { - "epoch": 0.89, - "learning_rate": 2.1966794380587487e-06, - "loss": 0.3345, + "epoch": 1.78, + "learning_rate": 2.192358951577077e-06, + "loss": 0.0834, "step": 16050 }, { - "epoch": 0.89, - "learning_rate": 2.191126658892776e-06, - "loss": 0.0896, + "epoch": 1.78, + "learning_rate": 2.1868058640604177e-06, + "loss": 0.1652, "step": 16055 }, { - "epoch": 0.89, - "learning_rate": 2.1855738797268035e-06, - "loss": 0.0967, + "epoch": 1.78, + "learning_rate": 2.1812527765437587e-06, + "loss": 0.2182, "step": 16060 }, { - "epoch": 0.89, - "learning_rate": 2.180021100560831e-06, - "loss": 0.1257, + "epoch": 1.78, + "learning_rate": 2.1756996890270994e-06, + "loss": 0.0748, "step": 16065 }, { - "epoch": 0.89, - "learning_rate": 2.1744683213948582e-06, - "loss": 0.119, + "epoch": 1.78, + "learning_rate": 2.17014660151044e-06, + "loss": 0.0713, "step": 16070 }, { - "epoch": 0.89, - "learning_rate": 2.1689155422288856e-06, - "loss": 0.1005, + "epoch": 1.78, + "learning_rate": 2.1645935139937806e-06, + "loss": 0.0682, "step": 16075 }, { - "epoch": 0.89, - "learning_rate": 2.163362763062913e-06, - "loss": 0.0514, + "epoch": 1.78, + "learning_rate": 2.1590404264771213e-06, + "loss": 0.0591, "step": 16080 }, { - "epoch": 0.89, - "learning_rate": 2.1578099838969404e-06, - "loss": 0.0864, + "epoch": 1.79, + "learning_rate": 2.1534873389604623e-06, + "loss": 0.029, "step": 16085 }, { - "epoch": 0.89, - "learning_rate": 2.152257204730968e-06, - "loss": 0.2015, + "epoch": 1.79, + "learning_rate": 2.147934251443803e-06, + "loss": 0.0431, "step": 16090 }, { - "epoch": 0.89, - "learning_rate": 2.1467044255649955e-06, - "loss": 0.1912, + "epoch": 1.79, + "learning_rate": 2.1423811639271436e-06, + "loss": 0.0603, "step": 16095 }, { - "epoch": 0.89, - "learning_rate": 2.141151646399023e-06, - "loss": 0.3672, + "epoch": 1.79, + "learning_rate": 2.1368280764104842e-06, + "loss": 0.1095, "step": 16100 }, { - "epoch": 0.89, - "learning_rate": 2.1355988672330503e-06, - "loss": 0.0839, + "epoch": 1.79, + "learning_rate": 2.131274988893825e-06, + "loss": 0.1472, "step": 16105 }, { - "epoch": 0.89, - "learning_rate": 2.130046088067078e-06, - "loss": 0.1216, + "epoch": 1.79, + "learning_rate": 2.125721901377166e-06, + "loss": 0.2512, "step": 16110 }, { - "epoch": 0.89, - "learning_rate": 2.124493308901105e-06, - "loss": 0.09, + "epoch": 1.79, + "learning_rate": 2.1201688138605066e-06, + "loss": 0.0478, "step": 16115 }, { - "epoch": 0.89, - "learning_rate": 2.1189405297351324e-06, - "loss": 0.1271, + "epoch": 1.79, + "learning_rate": 2.1146157263438472e-06, + "loss": 0.085, "step": 16120 }, { - "epoch": 0.89, - "learning_rate": 2.1133877505691598e-06, - "loss": 0.0858, + "epoch": 1.79, + "learning_rate": 2.1090626388271883e-06, + "loss": 0.0677, "step": 16125 }, { - "epoch": 0.9, - "learning_rate": 2.1078349714031876e-06, - "loss": 0.0677, + "epoch": 1.79, + "learning_rate": 2.103509551310529e-06, + "loss": 0.0685, "step": 16130 }, { - "epoch": 0.9, - "learning_rate": 2.102282192237215e-06, - "loss": 0.0322, + "epoch": 1.79, + "learning_rate": 2.0979564637938696e-06, + "loss": 0.0497, "step": 16135 }, { - "epoch": 0.9, - "learning_rate": 2.0967294130712423e-06, - "loss": 0.2141, + "epoch": 1.79, + "learning_rate": 2.09240337627721e-06, + "loss": 0.0172, "step": 16140 }, { - "epoch": 0.9, - "learning_rate": 2.0911766339052697e-06, - "loss": 0.1284, + "epoch": 1.79, + "learning_rate": 2.086850288760551e-06, + "loss": 0.0399, "step": 16145 }, { - "epoch": 0.9, - "learning_rate": 2.085623854739297e-06, - "loss": 0.3106, + "epoch": 1.79, + "learning_rate": 2.081297201243892e-06, + "loss": 0.0958, "step": 16150 }, { - "epoch": 0.9, - "learning_rate": 2.080071075573325e-06, - "loss": 0.1018, + "epoch": 1.79, + "learning_rate": 2.0757441137272325e-06, + "loss": 0.1884, "step": 16155 }, { - "epoch": 0.9, - "learning_rate": 2.074518296407352e-06, - "loss": 0.1304, + "epoch": 1.79, + "learning_rate": 2.070191026210573e-06, + "loss": 0.2624, "step": 16160 }, { - "epoch": 0.9, - "learning_rate": 2.0689655172413796e-06, - "loss": 0.1932, + "epoch": 1.79, + "learning_rate": 2.064637938693914e-06, + "loss": 0.0538, "step": 16165 }, { - "epoch": 0.9, - "learning_rate": 2.063412738075407e-06, - "loss": 0.1555, + "epoch": 1.79, + "learning_rate": 2.0590848511772545e-06, + "loss": 0.0522, "step": 16170 }, { - "epoch": 0.9, - "learning_rate": 2.0578599589094343e-06, - "loss": 0.0188, + "epoch": 1.8, + "learning_rate": 2.0535317636605955e-06, + "loss": 0.0489, "step": 16175 }, { - "epoch": 0.9, - "learning_rate": 2.0523071797434617e-06, - "loss": 0.0674, + "epoch": 1.8, + "learning_rate": 2.047978676143936e-06, + "loss": 0.0912, "step": 16180 }, { - "epoch": 0.9, - "learning_rate": 2.046754400577489e-06, - "loss": 0.0514, + "epoch": 1.8, + "learning_rate": 2.0424255886272772e-06, + "loss": 0.0451, "step": 16185 }, { - "epoch": 0.9, - "learning_rate": 2.0412016214115164e-06, - "loss": 0.1703, + "epoch": 1.8, + "learning_rate": 2.036872501110618e-06, + "loss": 0.0444, "step": 16190 }, { - "epoch": 0.9, - "learning_rate": 2.035648842245544e-06, - "loss": 0.1404, + "epoch": 1.8, + "learning_rate": 2.0313194135939585e-06, + "loss": 0.0648, "step": 16195 }, { - "epoch": 0.9, - "learning_rate": 2.0300960630795716e-06, - "loss": 0.375, + "epoch": 1.8, + "learning_rate": 2.025766326077299e-06, + "loss": 0.0708, "step": 16200 }, { - "epoch": 0.9, - "learning_rate": 2.024543283913599e-06, - "loss": 0.0987, + "epoch": 1.8, + "learning_rate": 2.0202132385606398e-06, + "loss": 0.1332, "step": 16205 }, { - "epoch": 0.9, - "learning_rate": 2.0189905047476264e-06, - "loss": 0.1256, + "epoch": 1.8, + "learning_rate": 2.0146601510439804e-06, + "loss": 0.2874, "step": 16210 }, { - "epoch": 0.9, - "learning_rate": 2.0134377255816537e-06, - "loss": 0.1031, + "epoch": 1.8, + "learning_rate": 2.0091070635273215e-06, + "loss": 0.0569, "step": 16215 }, { - "epoch": 0.9, - "learning_rate": 2.0078849464156815e-06, - "loss": 0.1061, + "epoch": 1.8, + "learning_rate": 2.003553976010662e-06, + "loss": 0.0601, "step": 16220 }, { - "epoch": 0.9, - "learning_rate": 2.002332167249709e-06, - "loss": 0.0968, + "epoch": 1.8, + "learning_rate": 1.9980008884940028e-06, + "loss": 0.0679, "step": 16225 }, { - "epoch": 0.9, - "learning_rate": 1.9967793880837363e-06, - "loss": 0.0698, + "epoch": 1.8, + "learning_rate": 1.9924478009773434e-06, + "loss": 0.0892, "step": 16230 }, { - "epoch": 0.9, - "learning_rate": 1.9912266089177632e-06, - "loss": 0.1025, + "epoch": 1.8, + "learning_rate": 1.986894713460684e-06, + "loss": 0.0253, "step": 16235 }, { - "epoch": 0.9, - "learning_rate": 1.985673829751791e-06, - "loss": 0.0985, + "epoch": 1.8, + "learning_rate": 1.981341625944025e-06, + "loss": 0.0429, "step": 16240 }, { - "epoch": 0.9, - "learning_rate": 1.9801210505858184e-06, - "loss": 0.1428, + "epoch": 1.8, + "learning_rate": 1.9757885384273657e-06, + "loss": 0.0915, "step": 16245 }, { - "epoch": 0.9, - "learning_rate": 1.9745682714198458e-06, - "loss": 0.4508, + "epoch": 1.8, + "learning_rate": 1.970235450910707e-06, + "loss": 0.0641, "step": 16250 }, { - "epoch": 0.9, - "learning_rate": 1.969015492253873e-06, - "loss": 0.0953, + "epoch": 1.8, + "learning_rate": 1.9646823633940474e-06, + "loss": 0.1163, "step": 16255 }, { - "epoch": 0.9, - "learning_rate": 1.9634627130879005e-06, - "loss": 0.094, + "epoch": 1.8, + "learning_rate": 1.959129275877388e-06, + "loss": 0.2137, "step": 16260 }, { - "epoch": 0.9, - "learning_rate": 1.9579099339219283e-06, - "loss": 0.1141, + "epoch": 1.81, + "learning_rate": 1.9535761883607287e-06, + "loss": 0.0449, "step": 16265 }, { - "epoch": 0.9, - "learning_rate": 1.9523571547559557e-06, - "loss": 0.0937, + "epoch": 1.81, + "learning_rate": 1.9480231008440694e-06, + "loss": 0.0721, "step": 16270 }, { - "epoch": 0.9, - "learning_rate": 1.946804375589983e-06, - "loss": 0.123, + "epoch": 1.81, + "learning_rate": 1.94247001332741e-06, + "loss": 0.0732, "step": 16275 }, { - "epoch": 0.9, - "learning_rate": 1.9412515964240104e-06, - "loss": 0.0577, + "epoch": 1.81, + "learning_rate": 1.936916925810751e-06, + "loss": 0.0931, "step": 16280 }, { - "epoch": 0.9, - "learning_rate": 1.9356988172580378e-06, - "loss": 0.0713, + "epoch": 1.81, + "learning_rate": 1.9313638382940917e-06, + "loss": 0.0452, "step": 16285 }, { - "epoch": 0.9, - "learning_rate": 1.930146038092065e-06, - "loss": 0.0989, + "epoch": 1.81, + "learning_rate": 1.9258107507774323e-06, + "loss": 0.061, "step": 16290 }, { - "epoch": 0.9, - "learning_rate": 1.9245932589260925e-06, - "loss": 0.091, + "epoch": 1.81, + "learning_rate": 1.920257663260773e-06, + "loss": 0.0606, "step": 16295 }, { - "epoch": 0.9, - "learning_rate": 1.91904047976012e-06, - "loss": 0.1646, + "epoch": 1.81, + "learning_rate": 1.9147045757441136e-06, + "loss": 0.0971, "step": 16300 }, { - "epoch": 0.9, - "learning_rate": 1.9134877005941473e-06, - "loss": 0.0726, + "epoch": 1.81, + "learning_rate": 1.9091514882274547e-06, + "loss": 0.1463, "step": 16305 }, { - "epoch": 0.91, - "learning_rate": 1.907934921428175e-06, - "loss": 0.0892, + "epoch": 1.81, + "learning_rate": 1.9035984007107955e-06, + "loss": 0.2243, "step": 16310 }, { - "epoch": 0.91, - "learning_rate": 1.9023821422622024e-06, - "loss": 0.0922, + "epoch": 1.81, + "learning_rate": 1.8980453131941362e-06, + "loss": 0.0595, "step": 16315 }, { - "epoch": 0.91, - "learning_rate": 1.8968293630962298e-06, - "loss": 0.0856, + "epoch": 1.81, + "learning_rate": 1.892492225677477e-06, + "loss": 0.0665, "step": 16320 }, { - "epoch": 0.91, - "learning_rate": 1.8912765839302572e-06, - "loss": 0.0763, + "epoch": 1.81, + "learning_rate": 1.8869391381608177e-06, + "loss": 0.0566, "step": 16325 }, { - "epoch": 0.91, - "learning_rate": 1.8857238047642848e-06, - "loss": 0.1137, + "epoch": 1.81, + "learning_rate": 1.8813860506441583e-06, + "loss": 0.0827, "step": 16330 }, { - "epoch": 0.91, - "learning_rate": 1.8801710255983121e-06, - "loss": 0.0307, + "epoch": 1.81, + "learning_rate": 1.8758329631274991e-06, + "loss": 0.0518, "step": 16335 }, { - "epoch": 0.91, - "learning_rate": 1.8746182464323395e-06, - "loss": 0.0938, + "epoch": 1.81, + "learning_rate": 1.8702798756108398e-06, + "loss": 0.0658, "step": 16340 }, { - "epoch": 0.91, - "learning_rate": 1.8690654672663669e-06, - "loss": 0.1645, + "epoch": 1.81, + "learning_rate": 1.8647267880941804e-06, + "loss": 0.0629, "step": 16345 }, { - "epoch": 0.91, - "learning_rate": 1.8635126881003943e-06, - "loss": 0.2979, + "epoch": 1.81, + "learning_rate": 1.8591737005775213e-06, + "loss": 0.0741, "step": 16350 }, { - "epoch": 0.91, - "learning_rate": 1.8579599089344218e-06, - "loss": 0.0975, + "epoch": 1.82, + "learning_rate": 1.853620613060862e-06, + "loss": 0.0842, "step": 16355 }, { - "epoch": 0.91, - "learning_rate": 1.8524071297684492e-06, - "loss": 0.0975, + "epoch": 1.82, + "learning_rate": 1.8480675255442028e-06, + "loss": 0.1819, "step": 16360 }, { - "epoch": 0.91, - "learning_rate": 1.8468543506024766e-06, - "loss": 0.1199, + "epoch": 1.82, + "learning_rate": 1.8425144380275434e-06, + "loss": 0.0578, "step": 16365 }, { - "epoch": 0.91, - "learning_rate": 1.841301571436504e-06, - "loss": 0.1449, + "epoch": 1.82, + "learning_rate": 1.836961350510884e-06, + "loss": 0.2826, "step": 16370 }, { - "epoch": 0.91, - "learning_rate": 1.8357487922705318e-06, - "loss": 0.1048, + "epoch": 1.82, + "learning_rate": 1.8314082629942249e-06, + "loss": 0.0595, "step": 16375 }, { - "epoch": 0.91, - "learning_rate": 1.830196013104559e-06, - "loss": 0.0386, + "epoch": 1.82, + "learning_rate": 1.8258551754775657e-06, + "loss": 0.0993, "step": 16380 }, { - "epoch": 0.91, - "learning_rate": 1.8246432339385863e-06, - "loss": 0.0495, + "epoch": 1.82, + "learning_rate": 1.8203020879609066e-06, + "loss": 0.0386, "step": 16385 }, { - "epoch": 0.91, - "learning_rate": 1.8190904547726137e-06, - "loss": 0.1614, + "epoch": 1.82, + "learning_rate": 1.8147490004442472e-06, + "loss": 0.0527, "step": 16390 }, { - "epoch": 0.91, - "learning_rate": 1.8135376756066415e-06, - "loss": 0.1107, + "epoch": 1.82, + "learning_rate": 1.8091959129275879e-06, + "loss": 0.0691, "step": 16395 }, { - "epoch": 0.91, - "learning_rate": 1.8079848964406688e-06, - "loss": 0.4405, + "epoch": 1.82, + "learning_rate": 1.8036428254109287e-06, + "loss": 0.0648, "step": 16400 }, { - "epoch": 0.91, - "learning_rate": 1.8024321172746962e-06, - "loss": 0.0963, + "epoch": 1.82, + "learning_rate": 1.7980897378942694e-06, + "loss": 0.1708, "step": 16405 }, { - "epoch": 0.91, - "learning_rate": 1.7968793381087236e-06, - "loss": 0.0958, + "epoch": 1.82, + "learning_rate": 1.79253665037761e-06, + "loss": 0.1871, "step": 16410 }, { - "epoch": 0.91, - "learning_rate": 1.791326558942751e-06, - "loss": 0.1276, + "epoch": 1.82, + "learning_rate": 1.7869835628609508e-06, + "loss": 0.0434, "step": 16415 }, { - "epoch": 0.91, - "learning_rate": 1.7857737797767785e-06, - "loss": 0.1165, + "epoch": 1.82, + "learning_rate": 1.7814304753442915e-06, + "loss": 0.0739, "step": 16420 }, { - "epoch": 0.91, - "learning_rate": 1.780221000610806e-06, - "loss": 0.0564, + "epoch": 1.82, + "learning_rate": 1.7758773878276323e-06, + "loss": 0.0589, "step": 16425 }, { - "epoch": 0.91, - "learning_rate": 1.7746682214448333e-06, - "loss": 0.0426, + "epoch": 1.82, + "learning_rate": 1.770324300310973e-06, + "loss": 0.0832, "step": 16430 }, { - "epoch": 0.91, - "learning_rate": 1.7691154422788606e-06, - "loss": 0.0672, + "epoch": 1.82, + "learning_rate": 1.7647712127943136e-06, + "loss": 0.0558, "step": 16435 }, { - "epoch": 0.91, - "learning_rate": 1.7635626631128882e-06, - "loss": 0.1156, + "epoch": 1.82, + "learning_rate": 1.7592181252776545e-06, + "loss": 0.0388, "step": 16440 }, { - "epoch": 0.91, - "learning_rate": 1.7580098839469156e-06, - "loss": 0.1564, + "epoch": 1.83, + "learning_rate": 1.7536650377609953e-06, + "loss": 0.0711, "step": 16445 }, { - "epoch": 0.91, - "learning_rate": 1.752457104780943e-06, - "loss": 0.3354, + "epoch": 1.83, + "learning_rate": 1.7481119502443362e-06, + "loss": 0.0703, "step": 16450 }, { - "epoch": 0.91, - "learning_rate": 1.7469043256149703e-06, - "loss": 0.0678, + "epoch": 1.83, + "learning_rate": 1.7425588627276768e-06, + "loss": 0.1202, "step": 16455 }, { - "epoch": 0.91, - "learning_rate": 1.7413515464489977e-06, - "loss": 0.098, + "epoch": 1.83, + "learning_rate": 1.7370057752110174e-06, + "loss": 0.2108, "step": 16460 }, { - "epoch": 0.91, - "learning_rate": 1.7357987672830253e-06, - "loss": 0.1186, + "epoch": 1.83, + "learning_rate": 1.7314526876943583e-06, + "loss": 0.0475, "step": 16465 }, { - "epoch": 0.91, - "learning_rate": 1.7302459881170527e-06, - "loss": 0.1185, + "epoch": 1.83, + "learning_rate": 1.725899600177699e-06, + "loss": 0.064, "step": 16470 }, { - "epoch": 0.91, - "learning_rate": 1.72469320895108e-06, - "loss": 0.145, + "epoch": 1.83, + "learning_rate": 1.7203465126610398e-06, + "loss": 0.0631, "step": 16475 }, { - "epoch": 0.91, - "learning_rate": 1.7191404297851074e-06, - "loss": 0.0653, + "epoch": 1.83, + "learning_rate": 1.7147934251443804e-06, + "loss": 0.0964, "step": 16480 }, { - "epoch": 0.91, - "learning_rate": 1.7135876506191352e-06, - "loss": 0.0533, + "epoch": 1.83, + "learning_rate": 1.709240337627721e-06, + "loss": 0.025, "step": 16485 }, { - "epoch": 0.92, - "learning_rate": 1.7080348714531626e-06, - "loss": 0.1078, + "epoch": 1.83, + "learning_rate": 1.703687250111062e-06, + "loss": 0.0707, "step": 16490 }, { - "epoch": 0.92, - "learning_rate": 1.70248209228719e-06, - "loss": 0.1496, + "epoch": 1.83, + "learning_rate": 1.6981341625944025e-06, + "loss": 0.0663, "step": 16495 }, { - "epoch": 0.92, - "learning_rate": 1.6969293131212171e-06, - "loss": 0.1952, + "epoch": 1.83, + "learning_rate": 1.6925810750777432e-06, + "loss": 0.1228, "step": 16500 }, { - "epoch": 0.92, - "learning_rate": 1.691376533955245e-06, - "loss": 0.0867, + "epoch": 1.83, + "learning_rate": 1.687027987561084e-06, + "loss": 0.1256, "step": 16505 }, { - "epoch": 0.92, - "learning_rate": 1.6858237547892723e-06, - "loss": 0.0948, + "epoch": 1.83, + "learning_rate": 1.6814749000444247e-06, + "loss": 0.2418, "step": 16510 }, { - "epoch": 0.92, - "learning_rate": 1.6802709756232997e-06, - "loss": 0.1396, + "epoch": 1.83, + "learning_rate": 1.6759218125277657e-06, + "loss": 0.0433, "step": 16515 }, { - "epoch": 0.92, - "learning_rate": 1.674718196457327e-06, - "loss": 0.0994, + "epoch": 1.83, + "learning_rate": 1.6703687250111064e-06, + "loss": 0.1052, "step": 16520 }, { - "epoch": 0.92, - "learning_rate": 1.6691654172913544e-06, - "loss": 0.1085, + "epoch": 1.83, + "learning_rate": 1.6648156374944472e-06, + "loss": 0.0842, "step": 16525 }, { - "epoch": 0.92, - "learning_rate": 1.663612638125382e-06, - "loss": 0.0681, + "epoch": 1.83, + "learning_rate": 1.6592625499777879e-06, + "loss": 0.0704, "step": 16530 }, { - "epoch": 0.92, - "learning_rate": 1.6580598589594094e-06, - "loss": 0.1306, + "epoch": 1.84, + "learning_rate": 1.6537094624611285e-06, + "loss": 0.0481, "step": 16535 }, { - "epoch": 0.92, - "learning_rate": 1.6525070797934367e-06, - "loss": 0.1745, + "epoch": 1.84, + "learning_rate": 1.6481563749444694e-06, + "loss": 0.0219, "step": 16540 }, { - "epoch": 0.92, - "learning_rate": 1.6469543006274641e-06, - "loss": 0.1977, + "epoch": 1.84, + "learning_rate": 1.64260328742781e-06, + "loss": 0.0998, "step": 16545 }, { - "epoch": 0.92, - "learning_rate": 1.6414015214614917e-06, - "loss": 0.2086, + "epoch": 1.84, + "learning_rate": 1.6370501999111506e-06, + "loss": 0.0682, "step": 16550 }, { - "epoch": 0.92, - "learning_rate": 1.635848742295519e-06, - "loss": 0.1545, + "epoch": 1.84, + "learning_rate": 1.6314971123944915e-06, + "loss": 0.1247, "step": 16555 }, { - "epoch": 0.92, - "learning_rate": 1.6302959631295464e-06, - "loss": 0.0889, + "epoch": 1.84, + "learning_rate": 1.6259440248778321e-06, + "loss": 0.1846, "step": 16560 }, { - "epoch": 0.92, - "learning_rate": 1.6247431839635738e-06, - "loss": 0.1235, + "epoch": 1.84, + "learning_rate": 1.6203909373611728e-06, + "loss": 0.0532, "step": 16565 }, { - "epoch": 0.92, - "learning_rate": 1.6191904047976012e-06, - "loss": 0.1358, + "epoch": 1.84, + "learning_rate": 1.6148378498445136e-06, + "loss": 0.0522, "step": 16570 }, { - "epoch": 0.92, - "learning_rate": 1.6136376256316288e-06, - "loss": 0.0324, + "epoch": 1.84, + "learning_rate": 1.6092847623278543e-06, + "loss": 0.0643, "step": 16575 }, { - "epoch": 0.92, - "learning_rate": 1.6080848464656561e-06, - "loss": 0.0737, + "epoch": 1.84, + "learning_rate": 1.6037316748111953e-06, + "loss": 0.1174, "step": 16580 }, { - "epoch": 0.92, - "learning_rate": 1.6025320672996835e-06, - "loss": 0.0405, + "epoch": 1.84, + "learning_rate": 1.598178587294536e-06, + "loss": 0.0303, "step": 16585 }, { - "epoch": 0.92, - "learning_rate": 1.5969792881337109e-06, - "loss": 0.1098, + "epoch": 1.84, + "learning_rate": 1.5926254997778768e-06, + "loss": 0.0597, "step": 16590 }, { - "epoch": 0.92, - "learning_rate": 1.5914265089677387e-06, - "loss": 0.2127, + "epoch": 1.84, + "learning_rate": 1.5870724122612174e-06, + "loss": 0.043, "step": 16595 }, { - "epoch": 0.92, - "learning_rate": 1.585873729801766e-06, - "loss": 0.3412, + "epoch": 1.84, + "learning_rate": 1.581519324744558e-06, + "loss": 0.0983, "step": 16600 }, { - "epoch": 0.92, - "learning_rate": 1.5803209506357934e-06, - "loss": 0.0887, + "epoch": 1.84, + "learning_rate": 1.575966237227899e-06, + "loss": 0.1792, "step": 16605 }, { - "epoch": 0.92, - "learning_rate": 1.5747681714698208e-06, - "loss": 0.0988, + "epoch": 1.84, + "learning_rate": 1.5704131497112396e-06, + "loss": 0.2726, "step": 16610 }, { - "epoch": 0.92, - "learning_rate": 1.5692153923038484e-06, - "loss": 0.0904, + "epoch": 1.84, + "learning_rate": 1.5648600621945802e-06, + "loss": 0.0432, "step": 16615 }, { - "epoch": 0.92, - "learning_rate": 1.5636626131378758e-06, - "loss": 0.1328, + "epoch": 1.84, + "learning_rate": 1.559306974677921e-06, + "loss": 0.0628, "step": 16620 }, { - "epoch": 0.92, - "learning_rate": 1.5581098339719031e-06, - "loss": 0.1044, + "epoch": 1.85, + "learning_rate": 1.5537538871612617e-06, + "loss": 0.0686, "step": 16625 }, { - "epoch": 0.92, - "learning_rate": 1.5525570548059305e-06, - "loss": 0.0428, + "epoch": 1.85, + "learning_rate": 1.5482007996446025e-06, + "loss": 0.0743, "step": 16630 }, { - "epoch": 0.92, - "learning_rate": 1.5470042756399579e-06, - "loss": 0.1033, + "epoch": 1.85, + "learning_rate": 1.5426477121279432e-06, + "loss": 0.0258, "step": 16635 }, { - "epoch": 0.92, - "learning_rate": 1.5414514964739855e-06, - "loss": 0.0904, + "epoch": 1.85, + "learning_rate": 1.5370946246112838e-06, + "loss": 0.0429, "step": 16640 }, { - "epoch": 0.92, - "learning_rate": 1.5358987173080128e-06, - "loss": 0.214, + "epoch": 1.85, + "learning_rate": 1.5315415370946247e-06, + "loss": 0.0762, "step": 16645 }, { - "epoch": 0.92, - "learning_rate": 1.5303459381420402e-06, - "loss": 0.2397, + "epoch": 1.85, + "learning_rate": 1.5259884495779655e-06, + "loss": 0.1121, "step": 16650 }, { - "epoch": 0.92, - "learning_rate": 1.5247931589760676e-06, - "loss": 0.0694, + "epoch": 1.85, + "learning_rate": 1.5204353620613064e-06, + "loss": 0.1533, "step": 16655 }, { - "epoch": 0.92, - "learning_rate": 1.5192403798100952e-06, - "loss": 0.1015, + "epoch": 1.85, + "learning_rate": 1.514882274544647e-06, + "loss": 0.2542, "step": 16660 }, { - "epoch": 0.92, - "learning_rate": 1.5136876006441225e-06, - "loss": 0.0569, + "epoch": 1.85, + "learning_rate": 1.5093291870279877e-06, + "loss": 0.0562, "step": 16665 }, { - "epoch": 0.93, - "learning_rate": 1.50813482147815e-06, - "loss": 0.1081, + "epoch": 1.85, + "learning_rate": 1.5037760995113285e-06, + "loss": 0.0469, "step": 16670 }, { - "epoch": 0.93, - "learning_rate": 1.5025820423121773e-06, - "loss": 0.0727, + "epoch": 1.85, + "learning_rate": 1.4982230119946691e-06, + "loss": 0.0685, "step": 16675 }, { - "epoch": 0.93, - "learning_rate": 1.4970292631462046e-06, - "loss": 0.0548, + "epoch": 1.85, + "learning_rate": 1.49266992447801e-06, + "loss": 0.0831, "step": 16680 }, { - "epoch": 0.93, - "learning_rate": 1.4914764839802322e-06, - "loss": 0.0535, + "epoch": 1.85, + "learning_rate": 1.4871168369613506e-06, + "loss": 0.029, "step": 16685 }, { - "epoch": 0.93, - "learning_rate": 1.4859237048142596e-06, - "loss": 0.1327, + "epoch": 1.85, + "learning_rate": 1.4815637494446913e-06, + "loss": 0.0232, "step": 16690 }, { - "epoch": 0.93, - "learning_rate": 1.480370925648287e-06, - "loss": 0.1224, + "epoch": 1.85, + "learning_rate": 1.4760106619280321e-06, + "loss": 0.0411, "step": 16695 }, { - "epoch": 0.93, - "learning_rate": 1.4748181464823143e-06, - "loss": 0.2078, + "epoch": 1.85, + "learning_rate": 1.4704575744113728e-06, + "loss": 0.0798, "step": 16700 }, { - "epoch": 0.93, - "learning_rate": 1.4692653673163421e-06, - "loss": 0.0955, + "epoch": 1.85, + "learning_rate": 1.4649044868947134e-06, + "loss": 0.1737, "step": 16705 }, { - "epoch": 0.93, - "learning_rate": 1.4637125881503695e-06, - "loss": 0.1178, + "epoch": 1.85, + "learning_rate": 1.4593513993780543e-06, + "loss": 0.2872, "step": 16710 }, { - "epoch": 0.93, - "learning_rate": 1.4581598089843969e-06, - "loss": 0.0971, + "epoch": 1.86, + "learning_rate": 1.4537983118613949e-06, + "loss": 0.0528, "step": 16715 }, { - "epoch": 0.93, - "learning_rate": 1.4526070298184243e-06, - "loss": 0.0929, + "epoch": 1.86, + "learning_rate": 1.448245224344736e-06, + "loss": 0.0541, "step": 16720 }, { - "epoch": 0.93, - "learning_rate": 1.4470542506524518e-06, - "loss": 0.0913, + "epoch": 1.86, + "learning_rate": 1.4426921368280766e-06, + "loss": 0.0532, "step": 16725 }, { - "epoch": 0.93, - "learning_rate": 1.4415014714864792e-06, - "loss": 0.0482, + "epoch": 1.86, + "learning_rate": 1.4371390493114174e-06, + "loss": 0.0761, "step": 16730 }, { - "epoch": 0.93, - "learning_rate": 1.4359486923205066e-06, - "loss": 0.1299, + "epoch": 1.86, + "learning_rate": 1.431585961794758e-06, + "loss": 0.054, "step": 16735 }, { - "epoch": 0.93, - "learning_rate": 1.430395913154534e-06, - "loss": 0.06, + "epoch": 1.86, + "learning_rate": 1.4260328742780987e-06, + "loss": 0.0266, "step": 16740 }, { - "epoch": 0.93, - "learning_rate": 1.4248431339885613e-06, - "loss": 0.0961, + "epoch": 1.86, + "learning_rate": 1.4204797867614396e-06, + "loss": 0.0609, "step": 16745 }, { - "epoch": 0.93, - "learning_rate": 1.419290354822589e-06, - "loss": 0.3137, + "epoch": 1.86, + "learning_rate": 1.4149266992447802e-06, + "loss": 0.1478, "step": 16750 }, { - "epoch": 0.93, - "learning_rate": 1.4137375756566163e-06, - "loss": 0.1318, + "epoch": 1.86, + "learning_rate": 1.4093736117281208e-06, + "loss": 0.1142, "step": 16755 }, { - "epoch": 0.93, - "learning_rate": 1.4081847964906437e-06, - "loss": 0.094, + "epoch": 1.86, + "learning_rate": 1.4038205242114617e-06, + "loss": 0.336, "step": 16760 }, { - "epoch": 0.93, - "learning_rate": 1.402632017324671e-06, - "loss": 0.0855, + "epoch": 1.86, + "learning_rate": 1.3982674366948023e-06, + "loss": 0.0549, "step": 16765 }, { - "epoch": 0.93, - "learning_rate": 1.3970792381586986e-06, - "loss": 0.1356, + "epoch": 1.86, + "learning_rate": 1.392714349178143e-06, + "loss": 0.0558, "step": 16770 }, { - "epoch": 0.93, - "learning_rate": 1.391526458992726e-06, - "loss": 0.0748, + "epoch": 1.86, + "learning_rate": 1.3871612616614838e-06, + "loss": 0.0541, "step": 16775 }, { - "epoch": 0.93, - "learning_rate": 1.3859736798267534e-06, - "loss": 0.0685, + "epoch": 1.86, + "learning_rate": 1.3816081741448245e-06, + "loss": 0.0753, "step": 16780 }, { - "epoch": 0.93, - "learning_rate": 1.3804209006607807e-06, - "loss": 0.055, + "epoch": 1.86, + "learning_rate": 1.3760550866281655e-06, + "loss": 0.0188, "step": 16785 }, { - "epoch": 0.93, - "learning_rate": 1.374868121494808e-06, - "loss": 0.0675, + "epoch": 1.86, + "learning_rate": 1.3705019991115062e-06, + "loss": 0.0573, "step": 16790 }, { - "epoch": 0.93, - "learning_rate": 1.3693153423288357e-06, - "loss": 0.1195, + "epoch": 1.86, + "learning_rate": 1.364948911594847e-06, + "loss": 0.0569, "step": 16795 }, { - "epoch": 0.93, - "learning_rate": 1.363762563162863e-06, - "loss": 0.3623, + "epoch": 1.86, + "learning_rate": 1.3593958240781877e-06, + "loss": 0.0927, "step": 16800 }, { - "epoch": 0.93, - "learning_rate": 1.3582097839968904e-06, - "loss": 0.1294, + "epoch": 1.87, + "learning_rate": 1.3538427365615283e-06, + "loss": 0.2221, "step": 16805 }, { - "epoch": 0.93, - "learning_rate": 1.3526570048309178e-06, - "loss": 0.1018, + "epoch": 1.87, + "learning_rate": 1.3482896490448691e-06, + "loss": 0.1996, "step": 16810 }, { - "epoch": 0.93, - "learning_rate": 1.3471042256649456e-06, - "loss": 0.0937, + "epoch": 1.87, + "learning_rate": 1.3427365615282098e-06, + "loss": 0.046, "step": 16815 }, { - "epoch": 0.93, - "learning_rate": 1.341551446498973e-06, - "loss": 0.0919, + "epoch": 1.87, + "learning_rate": 1.3371834740115504e-06, + "loss": 0.045, "step": 16820 }, { - "epoch": 0.93, - "learning_rate": 1.3359986673330003e-06, - "loss": 0.0653, + "epoch": 1.87, + "learning_rate": 1.3316303864948913e-06, + "loss": 0.052, "step": 16825 }, { - "epoch": 0.93, - "learning_rate": 1.3304458881670277e-06, - "loss": 0.0428, + "epoch": 1.87, + "learning_rate": 1.326077298978232e-06, + "loss": 0.079, "step": 16830 }, { - "epoch": 0.93, - "learning_rate": 1.324893109001055e-06, - "loss": 0.0621, + "epoch": 1.87, + "learning_rate": 1.3205242114615728e-06, + "loss": 0.0449, "step": 16835 }, { - "epoch": 0.93, - "learning_rate": 1.3193403298350827e-06, - "loss": 0.0869, + "epoch": 1.87, + "learning_rate": 1.3149711239449134e-06, + "loss": 0.0477, "step": 16840 }, { - "epoch": 0.93, - "learning_rate": 1.31378755066911e-06, - "loss": 0.1407, + "epoch": 1.87, + "learning_rate": 1.309418036428254e-06, + "loss": 0.0598, "step": 16845 }, { - "epoch": 0.94, - "learning_rate": 1.3082347715031374e-06, - "loss": 0.3887, + "epoch": 1.87, + "learning_rate": 1.3038649489115949e-06, + "loss": 0.0651, "step": 16850 }, { - "epoch": 0.94, - "learning_rate": 1.3026819923371648e-06, - "loss": 0.0927, + "epoch": 1.87, + "learning_rate": 1.2983118613949357e-06, + "loss": 0.106, "step": 16855 }, { - "epoch": 0.94, - "learning_rate": 1.2971292131711924e-06, - "loss": 0.0776, + "epoch": 1.87, + "learning_rate": 1.2927587738782766e-06, + "loss": 0.3549, "step": 16860 }, { - "epoch": 0.94, - "learning_rate": 1.2915764340052197e-06, - "loss": 0.0852, + "epoch": 1.87, + "learning_rate": 1.2872056863616172e-06, + "loss": 0.0437, "step": 16865 }, { - "epoch": 0.94, - "learning_rate": 1.2860236548392471e-06, - "loss": 0.1146, + "epoch": 1.87, + "learning_rate": 1.2816525988449579e-06, + "loss": 0.0762, "step": 16870 }, { - "epoch": 0.94, - "learning_rate": 1.2804708756732745e-06, - "loss": 0.1483, + "epoch": 1.87, + "learning_rate": 1.2760995113282987e-06, + "loss": 0.0547, "step": 16875 }, { - "epoch": 0.94, - "learning_rate": 1.274918096507302e-06, - "loss": 0.0354, + "epoch": 1.87, + "learning_rate": 1.2705464238116394e-06, + "loss": 0.1035, "step": 16880 }, { - "epoch": 0.94, - "learning_rate": 1.2693653173413294e-06, - "loss": 0.1186, + "epoch": 1.87, + "learning_rate": 1.2649933362949802e-06, + "loss": 0.0487, "step": 16885 }, { - "epoch": 0.94, - "learning_rate": 1.2638125381753568e-06, - "loss": 0.1239, + "epoch": 1.87, + "learning_rate": 1.2594402487783208e-06, + "loss": 0.0465, "step": 16890 }, { - "epoch": 0.94, - "learning_rate": 1.2582597590093842e-06, - "loss": 0.1191, + "epoch": 1.88, + "learning_rate": 1.2538871612616615e-06, + "loss": 0.0711, "step": 16895 }, { - "epoch": 0.94, - "learning_rate": 1.2527069798434116e-06, - "loss": 0.3428, + "epoch": 1.88, + "learning_rate": 1.2483340737450023e-06, + "loss": 0.1161, "step": 16900 }, { - "epoch": 0.94, - "learning_rate": 1.2471542006774391e-06, - "loss": 0.0996, + "epoch": 1.88, + "learning_rate": 1.2427809862283432e-06, + "loss": 0.1316, "step": 16905 }, { - "epoch": 0.94, - "learning_rate": 1.2416014215114665e-06, - "loss": 0.1159, + "epoch": 1.88, + "learning_rate": 1.2372278987116838e-06, + "loss": 0.1805, "step": 16910 }, { - "epoch": 0.94, - "learning_rate": 1.2360486423454939e-06, - "loss": 0.1285, + "epoch": 1.88, + "learning_rate": 1.2316748111950245e-06, + "loss": 0.0594, "step": 16915 }, { - "epoch": 0.94, - "learning_rate": 1.2304958631795215e-06, - "loss": 0.1126, + "epoch": 1.88, + "learning_rate": 1.2261217236783653e-06, + "loss": 0.0575, "step": 16920 }, - { - "epoch": 0.94, - "learning_rate": 1.2249430840135488e-06, - "loss": 0.0575, + { + "epoch": 1.88, + "learning_rate": 1.220568636161706e-06, + "loss": 0.0744, "step": 16925 }, { - "epoch": 0.94, - "learning_rate": 1.2193903048475764e-06, - "loss": 0.1218, + "epoch": 1.88, + "learning_rate": 1.2150155486450468e-06, + "loss": 0.0591, "step": 16930 }, { - "epoch": 0.94, - "learning_rate": 1.2138375256816038e-06, - "loss": 0.0996, + "epoch": 1.88, + "learning_rate": 1.2094624611283874e-06, + "loss": 0.0466, "step": 16935 }, { - "epoch": 0.94, - "learning_rate": 1.2082847465156312e-06, - "loss": 0.1539, + "epoch": 1.88, + "learning_rate": 1.2039093736117283e-06, + "loss": 0.0404, "step": 16940 }, { - "epoch": 0.94, - "learning_rate": 1.2027319673496585e-06, - "loss": 0.2066, + "epoch": 1.88, + "learning_rate": 1.198356286095069e-06, + "loss": 0.0875, "step": 16945 }, { - "epoch": 0.94, - "learning_rate": 1.197179188183686e-06, - "loss": 0.282, + "epoch": 1.88, + "learning_rate": 1.1928031985784098e-06, + "loss": 0.1315, "step": 16950 }, { - "epoch": 0.94, - "learning_rate": 1.1916264090177135e-06, - "loss": 0.0826, + "epoch": 1.88, + "learning_rate": 1.1872501110617504e-06, + "loss": 0.1174, "step": 16955 }, { - "epoch": 0.94, - "learning_rate": 1.1860736298517409e-06, - "loss": 0.0942, + "epoch": 1.88, + "learning_rate": 1.181697023545091e-06, + "loss": 0.2346, "step": 16960 }, { - "epoch": 0.94, - "learning_rate": 1.1805208506857685e-06, - "loss": 0.1454, + "epoch": 1.88, + "learning_rate": 1.176143936028432e-06, + "loss": 0.0677, "step": 16965 }, { - "epoch": 0.94, - "learning_rate": 1.1749680715197956e-06, - "loss": 0.0996, + "epoch": 1.88, + "learning_rate": 1.1705908485117728e-06, + "loss": 0.0564, "step": 16970 }, { - "epoch": 0.94, - "learning_rate": 1.1694152923538232e-06, - "loss": 0.1338, + "epoch": 1.88, + "learning_rate": 1.1650377609951134e-06, + "loss": 0.0629, "step": 16975 }, { - "epoch": 0.94, - "learning_rate": 1.1638625131878506e-06, - "loss": 0.0731, + "epoch": 1.88, + "learning_rate": 1.159484673478454e-06, + "loss": 0.087, "step": 16980 }, { - "epoch": 0.94, - "learning_rate": 1.1583097340218782e-06, - "loss": 0.0756, + "epoch": 1.89, + "learning_rate": 1.1539315859617949e-06, + "loss": 0.0499, "step": 16985 }, { - "epoch": 0.94, - "learning_rate": 1.1527569548559055e-06, - "loss": 0.1006, + "epoch": 1.89, + "learning_rate": 1.1483784984451355e-06, + "loss": 0.0328, "step": 16990 }, { - "epoch": 0.94, - "learning_rate": 1.147204175689933e-06, - "loss": 0.2047, + "epoch": 1.89, + "learning_rate": 1.1428254109284764e-06, + "loss": 0.056, "step": 16995 }, { - "epoch": 0.94, - "learning_rate": 1.1416513965239603e-06, - "loss": 0.3438, + "epoch": 1.89, + "learning_rate": 1.137272323411817e-06, + "loss": 0.0535, "step": 17000 }, { - "epoch": 0.94, - "learning_rate": 1.1360986173579877e-06, - "loss": 0.1061, + "epoch": 1.89, + "learning_rate": 1.1317192358951579e-06, + "loss": 0.1277, "step": 17005 }, { - "epoch": 0.94, - "learning_rate": 1.1305458381920152e-06, - "loss": 0.0866, + "epoch": 1.89, + "learning_rate": 1.1261661483784985e-06, + "loss": 0.2555, "step": 17010 }, { - "epoch": 0.94, - "learning_rate": 1.1249930590260426e-06, - "loss": 0.1258, + "epoch": 1.89, + "learning_rate": 1.1206130608618394e-06, + "loss": 0.0721, "step": 17015 }, { - "epoch": 0.94, - "learning_rate": 1.1194402798600702e-06, - "loss": 0.0998, + "epoch": 1.89, + "learning_rate": 1.11505997334518e-06, + "loss": 0.0513, "step": 17020 }, { - "epoch": 0.94, - "learning_rate": 1.1138875006940974e-06, - "loss": 0.1174, + "epoch": 1.89, + "learning_rate": 1.1095068858285206e-06, + "loss": 0.063, "step": 17025 }, { - "epoch": 0.95, - "learning_rate": 1.108334721528125e-06, - "loss": 0.1155, + "epoch": 1.89, + "learning_rate": 1.1039537983118615e-06, + "loss": 0.0801, "step": 17030 }, { - "epoch": 0.95, - "learning_rate": 1.1027819423621523e-06, - "loss": 0.1011, + "epoch": 1.89, + "learning_rate": 1.0984007107952021e-06, + "loss": 0.025, "step": 17035 }, { - "epoch": 0.95, - "learning_rate": 1.0972291631961799e-06, - "loss": 0.1669, + "epoch": 1.89, + "learning_rate": 1.092847623278543e-06, + "loss": 0.0329, "step": 17040 }, { - "epoch": 0.95, - "learning_rate": 1.0916763840302073e-06, - "loss": 0.2286, + "epoch": 1.89, + "learning_rate": 1.0872945357618838e-06, + "loss": 0.0712, "step": 17045 }, { - "epoch": 0.95, - "learning_rate": 1.0861236048642346e-06, - "loss": 0.3779, + "epoch": 1.89, + "learning_rate": 1.0817414482452245e-06, + "loss": 0.0956, "step": 17050 }, { - "epoch": 0.95, - "learning_rate": 1.080570825698262e-06, - "loss": 0.0815, + "epoch": 1.89, + "learning_rate": 1.076188360728565e-06, + "loss": 0.138, "step": 17055 }, { - "epoch": 0.95, - "learning_rate": 1.0750180465322894e-06, - "loss": 0.0931, + "epoch": 1.89, + "learning_rate": 1.070635273211906e-06, + "loss": 0.2153, "step": 17060 }, { - "epoch": 0.95, - "learning_rate": 1.069465267366317e-06, - "loss": 0.1096, + "epoch": 1.89, + "learning_rate": 1.0650821856952466e-06, + "loss": 0.0464, "step": 17065 }, { - "epoch": 0.95, - "learning_rate": 1.0639124882003443e-06, - "loss": 0.0962, + "epoch": 1.89, + "learning_rate": 1.0595290981785872e-06, + "loss": 0.0523, "step": 17070 }, { - "epoch": 0.95, - "learning_rate": 1.058359709034372e-06, - "loss": 0.0863, + "epoch": 1.9, + "learning_rate": 1.053976010661928e-06, + "loss": 0.0579, "step": 17075 }, { - "epoch": 0.95, - "learning_rate": 1.0528069298683993e-06, - "loss": 0.0333, + "epoch": 1.9, + "learning_rate": 1.048422923145269e-06, + "loss": 0.0793, "step": 17080 }, { - "epoch": 0.95, - "learning_rate": 1.0472541507024267e-06, - "loss": 0.0764, + "epoch": 1.9, + "learning_rate": 1.0428698356286096e-06, + "loss": 0.0534, "step": 17085 }, { - "epoch": 0.95, - "learning_rate": 1.041701371536454e-06, - "loss": 0.0922, + "epoch": 1.9, + "learning_rate": 1.0373167481119504e-06, + "loss": 0.0391, "step": 17090 }, { - "epoch": 0.95, - "learning_rate": 1.0361485923704816e-06, - "loss": 0.1658, + "epoch": 1.9, + "learning_rate": 1.031763660595291e-06, + "loss": 0.0708, "step": 17095 }, { - "epoch": 0.95, - "learning_rate": 1.030595813204509e-06, - "loss": 0.2404, + "epoch": 1.9, + "learning_rate": 1.0262105730786317e-06, + "loss": 0.0736, "step": 17100 }, { - "epoch": 0.95, - "learning_rate": 1.0250430340385364e-06, - "loss": 0.0939, + "epoch": 1.9, + "learning_rate": 1.0206574855619725e-06, + "loss": 0.1361, "step": 17105 }, { - "epoch": 0.95, - "learning_rate": 1.0194902548725637e-06, - "loss": 0.1102, + "epoch": 1.9, + "learning_rate": 1.0151043980453134e-06, + "loss": 0.1581, "step": 17110 }, { - "epoch": 0.95, - "learning_rate": 1.0139374757065911e-06, - "loss": 0.1249, + "epoch": 1.9, + "learning_rate": 1.009551310528654e-06, + "loss": 0.0619, "step": 17115 }, { - "epoch": 0.95, - "learning_rate": 1.0083846965406187e-06, - "loss": 0.1391, + "epoch": 1.9, + "learning_rate": 1.0039982230119947e-06, + "loss": 0.083, "step": 17120 }, { - "epoch": 0.95, - "learning_rate": 1.002831917374646e-06, - "loss": 0.1151, + "epoch": 1.9, + "learning_rate": 9.984451354953355e-07, + "loss": 0.0904, "step": 17125 }, { - "epoch": 0.95, - "learning_rate": 9.972791382086737e-07, - "loss": 0.0385, + "epoch": 1.9, + "learning_rate": 9.928920479786762e-07, + "loss": 0.0956, "step": 17130 }, { - "epoch": 0.95, - "learning_rate": 9.91726359042701e-07, - "loss": 0.1915, + "epoch": 1.9, + "learning_rate": 9.873389604620168e-07, + "loss": 0.0329, "step": 17135 }, { - "epoch": 0.95, - "learning_rate": 9.861735798767284e-07, - "loss": 0.195, + "epoch": 1.9, + "learning_rate": 9.817858729453579e-07, + "loss": 0.0485, "step": 17140 }, { - "epoch": 0.95, - "learning_rate": 9.806208007107558e-07, - "loss": 0.1971, + "epoch": 1.9, + "learning_rate": 9.762327854286985e-07, + "loss": 0.0824, "step": 17145 }, { - "epoch": 0.95, - "learning_rate": 9.750680215447834e-07, - "loss": 0.4488, + "epoch": 1.9, + "learning_rate": 9.706796979120391e-07, + "loss": 0.0728, "step": 17150 }, { - "epoch": 0.95, - "learning_rate": 9.695152423788107e-07, - "loss": 0.0723, + "epoch": 1.9, + "learning_rate": 9.6512661039538e-07, + "loss": 0.0969, "step": 17155 }, { - "epoch": 0.95, - "learning_rate": 9.63962463212838e-07, - "loss": 0.1043, + "epoch": 1.9, + "learning_rate": 9.595735228787206e-07, + "loss": 0.2431, "step": 17160 }, { - "epoch": 0.95, - "learning_rate": 9.584096840468655e-07, - "loss": 0.1299, + "epoch": 1.91, + "learning_rate": 9.540204353620613e-07, + "loss": 0.0669, "step": 17165 }, { - "epoch": 0.95, - "learning_rate": 9.528569048808929e-07, - "loss": 0.1067, + "epoch": 1.91, + "learning_rate": 9.48467347845402e-07, + "loss": 0.0446, "step": 17170 }, { - "epoch": 0.95, - "learning_rate": 9.473041257149204e-07, - "loss": 0.0326, + "epoch": 1.91, + "learning_rate": 9.429142603287429e-07, + "loss": 0.0779, "step": 17175 }, { - "epoch": 0.95, - "learning_rate": 9.417513465489478e-07, - "loss": 0.0517, + "epoch": 1.91, + "learning_rate": 9.373611728120836e-07, + "loss": 0.0687, "step": 17180 }, { - "epoch": 0.95, - "learning_rate": 9.361985673829753e-07, - "loss": 0.028, + "epoch": 1.91, + "learning_rate": 9.318080852954244e-07, + "loss": 0.062, "step": 17185 }, { - "epoch": 0.95, - "learning_rate": 9.306457882170026e-07, - "loss": 0.0894, + "epoch": 1.91, + "learning_rate": 9.262549977787651e-07, + "loss": 0.0501, "step": 17190 }, { - "epoch": 0.95, - "learning_rate": 9.250930090510301e-07, - "loss": 0.0953, + "epoch": 1.91, + "learning_rate": 9.207019102621057e-07, + "loss": 0.0608, "step": 17195 }, { - "epoch": 0.95, - "learning_rate": 9.195402298850575e-07, - "loss": 0.263, + "epoch": 1.91, + "learning_rate": 9.151488227454465e-07, + "loss": 0.0654, "step": 17200 }, { - "epoch": 0.95, - "learning_rate": 9.139874507190851e-07, - "loss": 0.1134, + "epoch": 1.91, + "learning_rate": 9.095957352287872e-07, + "loss": 0.1299, "step": 17205 }, { - "epoch": 0.96, - "learning_rate": 9.084346715531125e-07, - "loss": 0.0751, + "epoch": 1.91, + "learning_rate": 9.040426477121281e-07, + "loss": 0.2511, "step": 17210 }, { - "epoch": 0.96, - "learning_rate": 9.028818923871398e-07, - "loss": 0.0976, + "epoch": 1.91, + "learning_rate": 8.984895601954688e-07, + "loss": 0.0586, "step": 17215 }, { - "epoch": 0.96, - "learning_rate": 8.973291132211673e-07, - "loss": 0.1163, + "epoch": 1.91, + "learning_rate": 8.929364726788095e-07, + "loss": 0.0664, "step": 17220 }, { - "epoch": 0.96, - "learning_rate": 8.917763340551947e-07, - "loss": 0.0705, + "epoch": 1.91, + "learning_rate": 8.873833851621502e-07, + "loss": 0.0591, "step": 17225 }, { - "epoch": 0.96, - "learning_rate": 8.862235548892222e-07, - "loss": 0.0825, + "epoch": 1.91, + "learning_rate": 8.81830297645491e-07, + "loss": 0.0816, "step": 17230 }, { - "epoch": 0.96, - "learning_rate": 8.806707757232495e-07, - "loss": 0.0737, + "epoch": 1.91, + "learning_rate": 8.762772101288317e-07, + "loss": 0.0476, "step": 17235 }, { - "epoch": 0.96, - "learning_rate": 8.75117996557277e-07, - "loss": 0.1075, + "epoch": 1.91, + "learning_rate": 8.707241226121723e-07, + "loss": 0.0364, "step": 17240 }, { - "epoch": 0.96, - "learning_rate": 8.695652173913044e-07, - "loss": 0.1652, + "epoch": 1.91, + "learning_rate": 8.651710350955132e-07, + "loss": 0.0403, "step": 17245 }, { - "epoch": 0.96, - "learning_rate": 8.640124382253319e-07, - "loss": 0.2838, + "epoch": 1.91, + "learning_rate": 8.596179475788539e-07, + "loss": 0.1071, "step": 17250 }, { - "epoch": 0.96, - "learning_rate": 8.584596590593592e-07, - "loss": 0.0837, + "epoch": 1.92, + "learning_rate": 8.540648600621947e-07, + "loss": 0.097, "step": 17255 }, { - "epoch": 0.96, - "learning_rate": 8.529068798933868e-07, - "loss": 0.0872, + "epoch": 1.92, + "learning_rate": 8.485117725455354e-07, + "loss": 0.1938, "step": 17260 }, { - "epoch": 0.96, - "learning_rate": 8.473541007274142e-07, - "loss": 0.1167, + "epoch": 1.92, + "learning_rate": 8.429586850288761e-07, + "loss": 0.0511, "step": 17265 }, { - "epoch": 0.96, - "learning_rate": 8.418013215614416e-07, - "loss": 0.0544, + "epoch": 1.92, + "learning_rate": 8.374055975122168e-07, + "loss": 0.055, "step": 17270 }, { - "epoch": 0.96, - "learning_rate": 8.36248542395469e-07, - "loss": 0.0514, + "epoch": 1.92, + "learning_rate": 8.318525099955577e-07, + "loss": 0.0717, "step": 17275 }, { - "epoch": 0.96, - "learning_rate": 8.306957632294964e-07, - "loss": 0.0777, + "epoch": 1.92, + "learning_rate": 8.262994224788984e-07, + "loss": 0.0684, "step": 17280 }, { - "epoch": 0.96, - "learning_rate": 8.251429840635239e-07, - "loss": 0.166, + "epoch": 1.92, + "learning_rate": 8.207463349622391e-07, + "loss": 0.0777, "step": 17285 }, { - "epoch": 0.96, - "learning_rate": 8.195902048975513e-07, - "loss": 0.1412, + "epoch": 1.92, + "learning_rate": 8.151932474455798e-07, + "loss": 0.0308, "step": 17290 }, { - "epoch": 0.96, - "learning_rate": 8.140374257315787e-07, - "loss": 0.1778, + "epoch": 1.92, + "learning_rate": 8.096401599289205e-07, + "loss": 0.0899, "step": 17295 }, { - "epoch": 0.96, - "learning_rate": 8.084846465656061e-07, - "loss": 0.335, + "epoch": 1.92, + "learning_rate": 8.040870724122613e-07, + "loss": 0.1038, "step": 17300 }, { - "epoch": 0.96, - "learning_rate": 8.029318673996336e-07, - "loss": 0.0955, + "epoch": 1.92, + "learning_rate": 7.98533984895602e-07, + "loss": 0.1297, "step": 17305 }, { - "epoch": 0.96, - "learning_rate": 7.97379088233661e-07, - "loss": 0.112, + "epoch": 1.92, + "learning_rate": 7.929808973789429e-07, + "loss": 0.223, "step": 17310 }, { - "epoch": 0.96, - "learning_rate": 7.918263090676885e-07, - "loss": 0.0918, + "epoch": 1.92, + "learning_rate": 7.874278098622835e-07, + "loss": 0.0496, "step": 17315 }, { - "epoch": 0.96, - "learning_rate": 7.862735299017159e-07, - "loss": 0.0907, + "epoch": 1.92, + "learning_rate": 7.818747223456243e-07, + "loss": 0.0869, "step": 17320 }, { - "epoch": 0.96, - "learning_rate": 7.807207507357433e-07, - "loss": 0.0957, + "epoch": 1.92, + "learning_rate": 7.76321634828965e-07, + "loss": 0.0554, "step": 17325 }, { - "epoch": 0.96, - "learning_rate": 7.751679715697708e-07, - "loss": 0.0326, + "epoch": 1.92, + "learning_rate": 7.707685473123056e-07, + "loss": 0.0607, "step": 17330 }, { - "epoch": 0.96, - "learning_rate": 7.696151924037981e-07, - "loss": 0.1019, + "epoch": 1.92, + "learning_rate": 7.652154597956464e-07, + "loss": 0.0464, "step": 17335 }, { - "epoch": 0.96, - "learning_rate": 7.640624132378256e-07, - "loss": 0.1085, + "epoch": 1.92, + "learning_rate": 7.596623722789871e-07, + "loss": 0.0354, "step": 17340 }, { - "epoch": 0.96, - "learning_rate": 7.58509634071853e-07, - "loss": 0.1899, + "epoch": 1.93, + "learning_rate": 7.54109284762328e-07, + "loss": 0.0572, "step": 17345 }, { - "epoch": 0.96, - "learning_rate": 7.529568549058805e-07, - "loss": 0.3803, + "epoch": 1.93, + "learning_rate": 7.485561972456687e-07, + "loss": 0.0717, "step": 17350 }, { - "epoch": 0.96, - "learning_rate": 7.474040757399078e-07, - "loss": 0.0962, + "epoch": 1.93, + "learning_rate": 7.430031097290094e-07, + "loss": 0.2028, "step": 17355 }, { - "epoch": 0.96, - "learning_rate": 7.418512965739354e-07, - "loss": 0.0928, + "epoch": 1.93, + "learning_rate": 7.374500222123501e-07, + "loss": 0.3518, "step": 17360 }, { - "epoch": 0.96, - "learning_rate": 7.362985174079627e-07, - "loss": 0.0879, + "epoch": 1.93, + "learning_rate": 7.318969346956908e-07, + "loss": 0.061, "step": 17365 }, { - "epoch": 0.96, - "learning_rate": 7.307457382419903e-07, - "loss": 0.1282, + "epoch": 1.93, + "learning_rate": 7.263438471790316e-07, + "loss": 0.0577, "step": 17370 }, { - "epoch": 0.96, - "learning_rate": 7.251929590760176e-07, - "loss": 0.035, + "epoch": 1.93, + "learning_rate": 7.207907596623722e-07, + "loss": 0.0711, "step": 17375 }, { - "epoch": 0.96, - "learning_rate": 7.19640179910045e-07, - "loss": 0.0446, + "epoch": 1.93, + "learning_rate": 7.152376721457131e-07, + "loss": 0.0626, "step": 17380 }, { - "epoch": 0.96, - "learning_rate": 7.140874007440725e-07, - "loss": 0.1021, + "epoch": 1.93, + "learning_rate": 7.096845846290538e-07, + "loss": 0.0628, "step": 17385 }, { - "epoch": 0.97, - "learning_rate": 7.085346215780999e-07, - "loss": 0.1203, + "epoch": 1.93, + "learning_rate": 7.041314971123946e-07, + "loss": 0.0565, "step": 17390 }, { - "epoch": 0.97, - "learning_rate": 7.029818424121273e-07, - "loss": 0.1657, + "epoch": 1.93, + "learning_rate": 6.985784095957353e-07, + "loss": 0.0573, "step": 17395 }, { - "epoch": 0.97, - "learning_rate": 6.974290632461547e-07, - "loss": 0.3795, + "epoch": 1.93, + "learning_rate": 6.93025322079076e-07, + "loss": 0.0892, "step": 17400 }, { - "epoch": 0.97, - "learning_rate": 6.918762840801822e-07, - "loss": 0.1247, + "epoch": 1.93, + "learning_rate": 6.874722345624167e-07, + "loss": 0.106, "step": 17405 }, { - "epoch": 0.97, - "learning_rate": 6.863235049142096e-07, - "loss": 0.0748, + "epoch": 1.93, + "learning_rate": 6.819191470457574e-07, + "loss": 0.2682, "step": 17410 }, { - "epoch": 0.97, - "learning_rate": 6.807707257482372e-07, - "loss": 0.1084, + "epoch": 1.93, + "learning_rate": 6.763660595290983e-07, + "loss": 0.0555, "step": 17415 }, { - "epoch": 0.97, - "learning_rate": 6.752179465822645e-07, - "loss": 0.1146, + "epoch": 1.93, + "learning_rate": 6.70812972012439e-07, + "loss": 0.0701, "step": 17420 }, { - "epoch": 0.97, - "learning_rate": 6.69665167416292e-07, - "loss": 0.1076, + "epoch": 1.93, + "learning_rate": 6.652598844957797e-07, + "loss": 0.077, "step": 17425 }, { - "epoch": 0.97, - "learning_rate": 6.641123882503194e-07, - "loss": 0.0266, + "epoch": 1.93, + "learning_rate": 6.597067969791204e-07, + "loss": 0.0886, "step": 17430 }, { - "epoch": 0.97, - "learning_rate": 6.585596090843467e-07, - "loss": 0.077, + "epoch": 1.94, + "learning_rate": 6.541537094624612e-07, + "loss": 0.0371, "step": 17435 }, { - "epoch": 0.97, - "learning_rate": 6.530068299183742e-07, - "loss": 0.0936, + "epoch": 1.94, + "learning_rate": 6.486006219458019e-07, + "loss": 0.0482, "step": 17440 }, { - "epoch": 0.97, - "learning_rate": 6.474540507524016e-07, - "loss": 0.2206, + "epoch": 1.94, + "learning_rate": 6.430475344291428e-07, + "loss": 0.0555, "step": 17445 }, { - "epoch": 0.97, - "learning_rate": 6.419012715864291e-07, - "loss": 0.2803, + "epoch": 1.94, + "learning_rate": 6.374944469124834e-07, + "loss": 0.1553, "step": 17450 }, { - "epoch": 0.97, - "learning_rate": 6.363484924204564e-07, - "loss": 0.0762, + "epoch": 1.94, + "learning_rate": 6.319413593958241e-07, + "loss": 0.1197, "step": 17455 }, { - "epoch": 0.97, - "learning_rate": 6.307957132544839e-07, - "loss": 0.0861, + "epoch": 1.94, + "learning_rate": 6.263882718791649e-07, + "loss": 0.2412, "step": 17460 }, { - "epoch": 0.97, - "learning_rate": 6.252429340885113e-07, - "loss": 0.1163, + "epoch": 1.94, + "learning_rate": 6.208351843625056e-07, + "loss": 0.055, "step": 17465 }, { - "epoch": 0.97, - "learning_rate": 6.196901549225388e-07, - "loss": 0.09, + "epoch": 1.94, + "learning_rate": 6.152820968458464e-07, + "loss": 0.0713, "step": 17470 }, { - "epoch": 0.97, - "learning_rate": 6.141373757565663e-07, - "loss": 0.0313, + "epoch": 1.94, + "learning_rate": 6.097290093291871e-07, + "loss": 0.0752, "step": 17475 }, { - "epoch": 0.97, - "learning_rate": 6.085845965905936e-07, - "loss": 0.0558, + "epoch": 1.94, + "learning_rate": 6.041759218125278e-07, + "loss": 0.085, "step": 17480 }, { - "epoch": 0.97, - "learning_rate": 6.030318174246211e-07, - "loss": 0.1109, + "epoch": 1.94, + "learning_rate": 5.986228342958685e-07, + "loss": 0.034, "step": 17485 }, { - "epoch": 0.97, - "learning_rate": 5.974790382586485e-07, - "loss": 0.1782, + "epoch": 1.94, + "learning_rate": 5.930697467792094e-07, + "loss": 0.059, "step": 17490 }, { - "epoch": 0.97, - "learning_rate": 5.91926259092676e-07, - "loss": 0.2649, + "epoch": 1.94, + "learning_rate": 5.8751665926255e-07, + "loss": 0.0796, "step": 17495 }, { - "epoch": 0.97, - "learning_rate": 5.863734799267034e-07, - "loss": 0.2789, + "epoch": 1.94, + "learning_rate": 5.819635717458907e-07, + "loss": 0.0785, "step": 17500 }, { - "epoch": 0.97, - "learning_rate": 5.808207007607308e-07, - "loss": 0.0757, + "epoch": 1.94, + "learning_rate": 5.764104842292315e-07, + "loss": 0.1673, "step": 17505 }, { - "epoch": 0.97, - "learning_rate": 5.752679215947582e-07, - "loss": 0.0817, + "epoch": 1.94, + "learning_rate": 5.708573967125722e-07, + "loss": 0.2342, "step": 17510 }, { - "epoch": 0.97, - "learning_rate": 5.697151424287857e-07, - "loss": 0.1005, + "epoch": 1.94, + "learning_rate": 5.65304309195913e-07, + "loss": 0.0641, "step": 17515 }, { - "epoch": 0.97, - "learning_rate": 5.64162363262813e-07, - "loss": 0.1321, + "epoch": 1.94, + "learning_rate": 5.597512216792537e-07, + "loss": 0.0936, "step": 17520 }, { - "epoch": 0.97, - "learning_rate": 5.586095840968405e-07, - "loss": 0.1111, + "epoch": 1.95, + "learning_rate": 5.541981341625945e-07, + "loss": 0.0502, "step": 17525 }, { - "epoch": 0.97, - "learning_rate": 5.53056804930868e-07, - "loss": 0.0736, + "epoch": 1.95, + "learning_rate": 5.486450466459352e-07, + "loss": 0.0633, "step": 17530 }, { - "epoch": 0.97, - "learning_rate": 5.475040257648954e-07, - "loss": 0.0793, + "epoch": 1.95, + "learning_rate": 5.430919591292758e-07, + "loss": 0.0599, "step": 17535 }, { - "epoch": 0.97, - "learning_rate": 5.419512465989228e-07, - "loss": 0.0731, + "epoch": 1.95, + "learning_rate": 5.375388716126167e-07, + "loss": 0.0471, "step": 17540 }, { - "epoch": 0.97, - "learning_rate": 5.363984674329502e-07, - "loss": 0.1968, + "epoch": 1.95, + "learning_rate": 5.319857840959574e-07, + "loss": 0.052, "step": 17545 }, { - "epoch": 0.97, - "learning_rate": 5.308456882669777e-07, - "loss": 0.31, + "epoch": 1.95, + "learning_rate": 5.264326965792981e-07, + "loss": 0.1048, "step": 17550 }, { - "epoch": 0.97, - "learning_rate": 5.252929091010052e-07, - "loss": 0.0805, + "epoch": 1.95, + "learning_rate": 5.208796090626389e-07, + "loss": 0.1494, "step": 17555 }, { - "epoch": 0.97, - "learning_rate": 5.197401299350325e-07, - "loss": 0.1076, + "epoch": 1.95, + "learning_rate": 5.164371390493115e-07, + "loss": 0.2208, "step": 17560 }, { - "epoch": 0.97, - "learning_rate": 5.141873507690599e-07, - "loss": 0.2134, + "epoch": 1.95, + "learning_rate": 5.108840515326522e-07, + "loss": 0.0672, "step": 17565 }, { - "epoch": 0.98, - "learning_rate": 5.086345716030874e-07, - "loss": 0.1235, + "epoch": 1.95, + "learning_rate": 5.053309640159929e-07, + "loss": 0.0598, "step": 17570 }, { - "epoch": 0.98, - "learning_rate": 5.030817924371148e-07, - "loss": 0.0414, + "epoch": 1.95, + "learning_rate": 4.997778764993336e-07, + "loss": 0.0659, "step": 17575 }, { - "epoch": 0.98, - "learning_rate": 4.975290132711422e-07, - "loss": 0.0814, + "epoch": 1.95, + "learning_rate": 4.942247889826745e-07, + "loss": 0.0703, "step": 17580 }, { - "epoch": 0.98, - "learning_rate": 4.919762341051697e-07, - "loss": 0.1218, + "epoch": 1.95, + "learning_rate": 4.886717014660151e-07, + "loss": 0.0651, "step": 17585 }, { - "epoch": 0.98, - "learning_rate": 4.864234549391971e-07, - "loss": 0.1068, + "epoch": 1.95, + "learning_rate": 4.831186139493558e-07, + "loss": 0.082, "step": 17590 }, { - "epoch": 0.98, - "learning_rate": 4.808706757732246e-07, - "loss": 0.1727, + "epoch": 1.95, + "learning_rate": 4.775655264326966e-07, + "loss": 0.0468, "step": 17595 }, { - "epoch": 0.98, - "learning_rate": 4.75317896607252e-07, - "loss": 0.5324, + "epoch": 1.95, + "learning_rate": 4.7201243891603734e-07, + "loss": 0.0696, "step": 17600 }, { - "epoch": 0.98, - "learning_rate": 4.697651174412794e-07, - "loss": 0.0784, + "epoch": 1.95, + "learning_rate": 4.664593513993781e-07, + "loss": 0.1671, "step": 17605 }, { - "epoch": 0.98, - "learning_rate": 4.6421233827530684e-07, - "loss": 0.0668, + "epoch": 1.95, + "learning_rate": 4.609062638827189e-07, + "loss": 0.2603, "step": 17610 }, { - "epoch": 0.98, - "learning_rate": 4.586595591093342e-07, - "loss": 0.0939, + "epoch": 1.96, + "learning_rate": 4.5535317636605957e-07, + "loss": 0.0437, "step": 17615 }, { - "epoch": 0.98, - "learning_rate": 4.5310677994336164e-07, - "loss": 0.0938, + "epoch": 1.96, + "learning_rate": 4.498000888494003e-07, + "loss": 0.091, "step": 17620 }, { - "epoch": 0.98, - "learning_rate": 4.475540007773891e-07, - "loss": 0.0703, + "epoch": 1.96, + "learning_rate": 4.44247001332741e-07, + "loss": 0.0633, "step": 17625 }, { - "epoch": 0.98, - "learning_rate": 4.4200122161141654e-07, - "loss": 0.0535, + "epoch": 1.96, + "learning_rate": 4.386939138160818e-07, + "loss": 0.0697, "step": 17630 }, { - "epoch": 0.98, - "learning_rate": 4.3644844244544397e-07, - "loss": 0.1055, + "epoch": 1.96, + "learning_rate": 4.331408262994225e-07, + "loss": 0.0647, "step": 17635 }, { - "epoch": 0.98, - "learning_rate": 4.308956632794714e-07, - "loss": 0.1823, + "epoch": 1.96, + "learning_rate": 4.2758773878276324e-07, + "loss": 0.0519, "step": 17640 }, { - "epoch": 0.98, - "learning_rate": 4.2534288411349887e-07, - "loss": 0.1584, + "epoch": 1.96, + "learning_rate": 4.2203465126610404e-07, + "loss": 0.0593, "step": 17645 }, { - "epoch": 0.98, - "learning_rate": 4.197901049475263e-07, - "loss": 0.3552, + "epoch": 1.96, + "learning_rate": 4.1648156374944473e-07, + "loss": 0.072, "step": 17650 }, { - "epoch": 0.98, - "learning_rate": 4.142373257815537e-07, - "loss": 0.0609, + "epoch": 1.96, + "learning_rate": 4.109284762327854e-07, + "loss": 0.0744, "step": 17655 }, { - "epoch": 0.98, - "learning_rate": 4.0868454661558115e-07, - "loss": 0.1043, + "epoch": 1.96, + "learning_rate": 4.0537538871612617e-07, + "loss": 0.1935, "step": 17660 }, { - "epoch": 0.98, - "learning_rate": 4.0313176744960857e-07, - "loss": 0.0898, + "epoch": 1.96, + "learning_rate": 3.9982230119946697e-07, + "loss": 0.0547, "step": 17665 }, { - "epoch": 0.98, - "learning_rate": 3.9757898828363594e-07, - "loss": 0.1073, + "epoch": 1.96, + "learning_rate": 3.9426921368280766e-07, + "loss": 0.069, "step": 17670 }, { - "epoch": 0.98, - "learning_rate": 3.920262091176634e-07, - "loss": 0.0371, + "epoch": 1.96, + "learning_rate": 3.887161261661484e-07, + "loss": 0.0513, "step": 17675 }, { - "epoch": 0.98, - "learning_rate": 3.8647342995169085e-07, - "loss": 0.1446, + "epoch": 1.96, + "learning_rate": 3.8316303864948915e-07, + "loss": 0.0775, "step": 17680 }, { - "epoch": 0.98, - "learning_rate": 3.8092065078571827e-07, - "loss": 0.1384, + "epoch": 1.96, + "learning_rate": 3.776099511328299e-07, + "loss": 0.0484, "step": 17685 }, { - "epoch": 0.98, - "learning_rate": 3.753678716197457e-07, - "loss": 0.1071, + "epoch": 1.96, + "learning_rate": 3.720568636161706e-07, + "loss": 0.0421, "step": 17690 }, { - "epoch": 0.98, - "learning_rate": 3.698150924537731e-07, - "loss": 0.1018, + "epoch": 1.96, + "learning_rate": 3.665037760995114e-07, + "loss": 0.0791, "step": 17695 }, { - "epoch": 0.98, - "learning_rate": 3.642623132878006e-07, - "loss": 0.4557, + "epoch": 1.96, + "learning_rate": 3.609506885828521e-07, + "loss": 0.0946, "step": 17700 }, { - "epoch": 0.98, - "learning_rate": 3.58709534121828e-07, - "loss": 0.0818, + "epoch": 1.97, + "learning_rate": 3.553976010661928e-07, + "loss": 0.1163, "step": 17705 }, { - "epoch": 0.98, - "learning_rate": 3.5315675495585545e-07, - "loss": 0.0919, + "epoch": 1.97, + "learning_rate": 3.4984451354953356e-07, + "loss": 0.1559, "step": 17710 }, { - "epoch": 0.98, - "learning_rate": 3.476039757898829e-07, - "loss": 0.0948, + "epoch": 1.97, + "learning_rate": 3.442914260328743e-07, + "loss": 0.0571, "step": 17715 }, { - "epoch": 0.98, - "learning_rate": 3.4205119662391035e-07, - "loss": 0.1088, + "epoch": 1.97, + "learning_rate": 3.3873833851621505e-07, + "loss": 0.0554, "step": 17720 }, { - "epoch": 0.98, - "learning_rate": 3.3649841745793767e-07, - "loss": 0.0761, + "epoch": 1.97, + "learning_rate": 3.3318525099955574e-07, + "loss": 0.0811, "step": 17725 }, { - "epoch": 0.98, - "learning_rate": 3.3094563829196515e-07, - "loss": 0.1205, + "epoch": 1.97, + "learning_rate": 3.2763216348289654e-07, + "loss": 0.0936, "step": 17730 }, { - "epoch": 0.98, - "learning_rate": 3.253928591259926e-07, - "loss": 0.0723, + "epoch": 1.97, + "learning_rate": 3.220790759662373e-07, + "loss": 0.0368, "step": 17735 }, { - "epoch": 0.98, - "learning_rate": 3.1984007996002e-07, - "loss": 0.079, + "epoch": 1.97, + "learning_rate": 3.16525988449578e-07, + "loss": 0.0344, "step": 17740 }, { - "epoch": 0.98, - "learning_rate": 3.1428730079404743e-07, - "loss": 0.2108, + "epoch": 1.97, + "learning_rate": 3.109729009329187e-07, + "loss": 0.0587, "step": 17745 }, { - "epoch": 0.99, - "learning_rate": 3.087345216280749e-07, - "loss": 0.2737, + "epoch": 1.97, + "learning_rate": 3.0541981341625947e-07, + "loss": 0.0684, "step": 17750 }, { - "epoch": 0.99, - "learning_rate": 3.0318174246210233e-07, - "loss": 0.1012, + "epoch": 1.97, + "learning_rate": 2.998667258996002e-07, + "loss": 0.0947, "step": 17755 }, { - "epoch": 0.99, - "learning_rate": 2.9762896329612976e-07, - "loss": 0.0872, + "epoch": 1.97, + "learning_rate": 2.9431363838294096e-07, + "loss": 0.1762, "step": 17760 }, { - "epoch": 0.99, - "learning_rate": 2.920761841301572e-07, - "loss": 0.1619, + "epoch": 1.97, + "learning_rate": 2.8876055086628165e-07, + "loss": 0.0766, "step": 17765 }, { - "epoch": 0.99, - "learning_rate": 2.865234049641846e-07, - "loss": 0.1314, + "epoch": 1.97, + "learning_rate": 2.832074633496224e-07, + "loss": 0.0565, "step": 17770 }, { - "epoch": 0.99, - "learning_rate": 2.8097062579821203e-07, - "loss": 0.0481, + "epoch": 1.97, + "learning_rate": 2.776543758329632e-07, + "loss": 0.0718, "step": 17775 }, { - "epoch": 0.99, - "learning_rate": 2.7541784663223946e-07, - "loss": 0.0624, + "epoch": 1.97, + "learning_rate": 2.721012883163039e-07, + "loss": 0.0708, "step": 17780 }, { - "epoch": 0.99, - "learning_rate": 2.698650674662669e-07, - "loss": 0.1149, + "epoch": 1.97, + "learning_rate": 2.6654820079964463e-07, + "loss": 0.0353, "step": 17785 }, { - "epoch": 0.99, - "learning_rate": 2.643122883002943e-07, - "loss": 0.0606, + "epoch": 1.97, + "learning_rate": 2.6099511328298537e-07, + "loss": 0.0346, "step": 17790 }, { - "epoch": 0.99, - "learning_rate": 2.5875950913432173e-07, - "loss": 0.0961, + "epoch": 1.98, + "learning_rate": 2.554420257663261e-07, + "loss": 0.0675, "step": 17795 }, { - "epoch": 0.99, - "learning_rate": 2.5320672996834916e-07, - "loss": 0.2126, + "epoch": 1.98, + "learning_rate": 2.498889382496668e-07, + "loss": 0.0794, "step": 17800 }, { - "epoch": 0.99, - "learning_rate": 2.4765395080237663e-07, - "loss": 0.0577, + "epoch": 1.98, + "learning_rate": 2.4433585073300755e-07, + "loss": 0.1368, "step": 17805 }, { - "epoch": 0.99, - "learning_rate": 2.4210117163640406e-07, - "loss": 0.0961, + "epoch": 1.98, + "learning_rate": 2.387827632163483e-07, + "loss": 0.1639, "step": 17810 }, { - "epoch": 0.99, - "learning_rate": 2.3654839247043149e-07, - "loss": 0.0998, + "epoch": 1.98, + "learning_rate": 2.3322967569968904e-07, + "loss": 0.0483, "step": 17815 }, { - "epoch": 0.99, - "learning_rate": 2.3099561330445888e-07, - "loss": 0.108, + "epoch": 1.98, + "learning_rate": 2.2767658818302979e-07, + "loss": 0.0509, "step": 17820 }, { - "epoch": 0.99, - "learning_rate": 2.2544283413848634e-07, - "loss": 0.0743, + "epoch": 1.98, + "learning_rate": 2.221235006663705e-07, + "loss": 0.0627, "step": 17825 }, { - "epoch": 0.99, - "learning_rate": 2.1989005497251376e-07, - "loss": 0.0361, + "epoch": 1.98, + "learning_rate": 2.1657041314971125e-07, + "loss": 0.0716, "step": 17830 }, { - "epoch": 0.99, - "learning_rate": 2.1433727580654119e-07, - "loss": 0.1311, + "epoch": 1.98, + "learning_rate": 2.1101732563305202e-07, + "loss": 0.0484, "step": 17835 }, { - "epoch": 0.99, - "learning_rate": 2.0878449664056864e-07, - "loss": 0.1033, + "epoch": 1.98, + "learning_rate": 2.054642381163927e-07, + "loss": 0.0651, "step": 17840 }, { - "epoch": 0.99, - "learning_rate": 2.0323171747459604e-07, - "loss": 0.1616, + "epoch": 1.98, + "learning_rate": 1.9991115059973348e-07, + "loss": 0.0576, "step": 17845 }, { - "epoch": 0.99, - "learning_rate": 1.9767893830862346e-07, - "loss": 0.3498, + "epoch": 1.98, + "learning_rate": 1.943580630830742e-07, + "loss": 0.0798, "step": 17850 }, { - "epoch": 0.99, - "learning_rate": 1.9212615914265091e-07, - "loss": 0.0685, + "epoch": 1.98, + "learning_rate": 1.8880497556641495e-07, + "loss": 0.1742, "step": 17855 }, { - "epoch": 0.99, - "learning_rate": 1.8657337997667834e-07, - "loss": 0.1194, + "epoch": 1.98, + "learning_rate": 1.832518880497557e-07, + "loss": 0.2186, "step": 17860 }, { - "epoch": 0.99, - "learning_rate": 1.810206008107058e-07, - "loss": 0.1161, + "epoch": 1.98, + "learning_rate": 1.776988005330964e-07, + "loss": 0.0784, "step": 17865 }, { - "epoch": 0.99, - "learning_rate": 1.754678216447332e-07, - "loss": 0.1092, + "epoch": 1.98, + "learning_rate": 1.7214571301643715e-07, + "loss": 0.0432, "step": 17870 }, { - "epoch": 0.99, - "learning_rate": 1.6991504247876061e-07, - "loss": 0.0634, + "epoch": 1.98, + "learning_rate": 1.6659262549977787e-07, + "loss": 0.0681, "step": 17875 }, { - "epoch": 0.99, - "learning_rate": 1.6436226331278807e-07, - "loss": 0.0525, + "epoch": 1.98, + "learning_rate": 1.6103953798311864e-07, + "loss": 0.0903, "step": 17880 }, { - "epoch": 0.99, - "learning_rate": 1.588094841468155e-07, - "loss": 0.0914, + "epoch": 1.99, + "learning_rate": 1.5548645046645936e-07, + "loss": 0.0224, "step": 17885 }, { - "epoch": 0.99, - "learning_rate": 1.5325670498084292e-07, - "loss": 0.1541, + "epoch": 1.99, + "learning_rate": 1.499333629498001e-07, + "loss": 0.0345, "step": 17890 }, { - "epoch": 0.99, - "learning_rate": 1.4770392581487034e-07, - "loss": 0.283, + "epoch": 1.99, + "learning_rate": 1.4438027543314082e-07, + "loss": 0.0822, "step": 17895 }, { - "epoch": 0.99, - "learning_rate": 1.421511466488978e-07, - "loss": 0.3174, + "epoch": 1.99, + "learning_rate": 1.388271879164816e-07, + "loss": 0.094, "step": 17900 }, { - "epoch": 0.99, - "learning_rate": 1.3659836748292522e-07, - "loss": 0.0833, + "epoch": 1.99, + "learning_rate": 1.3327410039982231e-07, + "loss": 0.1584, "step": 17905 }, { - "epoch": 0.99, - "learning_rate": 1.3104558831695264e-07, - "loss": 0.0913, + "epoch": 1.99, + "learning_rate": 1.2772101288316306e-07, + "loss": 0.1863, "step": 17910 }, { - "epoch": 0.99, - "learning_rate": 1.254928091509801e-07, - "loss": 0.1119, + "epoch": 1.99, + "learning_rate": 1.2216792536650378e-07, + "loss": 0.049, "step": 17915 }, { - "epoch": 0.99, - "learning_rate": 1.199400299850075e-07, - "loss": 0.0809, + "epoch": 1.99, + "learning_rate": 1.1661483784984452e-07, + "loss": 0.0551, "step": 17920 }, { - "epoch": 0.99, - "learning_rate": 1.1438725081903493e-07, - "loss": 0.075, + "epoch": 1.99, + "learning_rate": 1.1106175033318525e-07, + "loss": 0.0665, "step": 17925 }, { - "epoch": 1.0, - "learning_rate": 1.0883447165306236e-07, - "loss": 0.0956, + "epoch": 1.99, + "learning_rate": 1.0550866281652601e-07, + "loss": 0.068, "step": 17930 }, { - "epoch": 1.0, - "learning_rate": 1.032816924870898e-07, - "loss": 0.0297, + "epoch": 1.99, + "learning_rate": 9.995557529986674e-08, + "loss": 0.0549, "step": 17935 }, { - "epoch": 1.0, - "learning_rate": 9.772891332111723e-08, - "loss": 0.148, + "epoch": 1.99, + "learning_rate": 9.440248778320747e-08, + "loss": 0.0376, "step": 17940 }, { - "epoch": 1.0, - "learning_rate": 9.217613415514465e-08, - "loss": 0.0918, + "epoch": 1.99, + "learning_rate": 8.88494002665482e-08, + "loss": 0.0756, "step": 17945 }, { - "epoch": 1.0, - "learning_rate": 8.662335498917208e-08, - "loss": 0.3377, + "epoch": 1.99, + "learning_rate": 8.329631274988894e-08, + "loss": 0.1298, "step": 17950 }, { - "epoch": 1.0, - "learning_rate": 8.107057582319952e-08, - "loss": 0.0697, + "epoch": 1.99, + "learning_rate": 7.774322523322968e-08, + "loss": 0.1729, "step": 17955 }, { - "epoch": 1.0, - "learning_rate": 7.551779665722695e-08, - "loss": 0.0934, + "epoch": 1.99, + "learning_rate": 7.219013771657041e-08, + "loss": 0.2417, "step": 17960 }, { - "epoch": 1.0, - "learning_rate": 6.996501749125439e-08, - "loss": 0.0987, + "epoch": 1.99, + "learning_rate": 6.663705019991116e-08, + "loss": 0.0607, "step": 17965 }, { - "epoch": 1.0, - "learning_rate": 6.441223832528181e-08, - "loss": 0.0754, + "epoch": 1.99, + "learning_rate": 6.108396268325189e-08, + "loss": 0.0679, "step": 17970 }, { - "epoch": 1.0, - "learning_rate": 5.885945915930924e-08, - "loss": 0.0613, + "epoch": 2.0, + "learning_rate": 5.5530875166592626e-08, + "loss": 0.0762, "step": 17975 }, { - "epoch": 1.0, - "learning_rate": 5.3306679993336676e-08, - "loss": 0.016, + "epoch": 2.0, + "learning_rate": 4.997778764993337e-08, + "loss": 0.0721, "step": 17980 }, { - "epoch": 1.0, - "learning_rate": 4.77539008273641e-08, - "loss": 0.086, + "epoch": 2.0, + "learning_rate": 4.44247001332741e-08, + "loss": 0.0554, "step": 17985 }, { - "epoch": 1.0, - "learning_rate": 4.220112166139153e-08, - "loss": 0.1172, + "epoch": 2.0, + "learning_rate": 3.887161261661484e-08, + "loss": 0.042, "step": 17990 }, { - "epoch": 1.0, - "learning_rate": 3.6648342495418965e-08, - "loss": 0.1962, + "epoch": 2.0, + "learning_rate": 3.331852509995558e-08, + "loss": 0.0646, "step": 17995 }, { - "epoch": 1.0, - "learning_rate": 3.109556332944639e-08, - "loss": 0.2095, + "epoch": 2.0, + "learning_rate": 2.7765437583296313e-08, + "loss": 0.0696, "step": 18000 }, { - "epoch": 1.0, - "eval_loss": 0.039903707802295685, - "eval_runtime": 1397.6965, - "eval_samples_per_second": 11.46, - "eval_steps_per_second": 2.865, - "eval_wer": 0.1074828705373242, + "epoch": 2.0, + "eval_loss": 0.0309030469506979, + "eval_runtime": 1852.9124, + "eval_samples_per_second": 8.645, + "eval_steps_per_second": 2.161, + "eval_wer": 0.06855391272989542, "step": 18000 } ], - "max_steps": 18019, - "num_train_epochs": 1, - "total_flos": 4.4943399434056515e+18, + "max_steps": 18018, + "num_train_epochs": 2, + "total_flos": 8.926067754739164e+18, "trial_name": null, "trial_params": null }