{ "config": { "vocab_size": 32000, "d_model": 1280, "n_layers": 23, "n_heads": 20, "d_ff": 3456, "max_seq_len": 512, "rope_theta": 10000.0, "rope_pct": 0.25, "dropout": 0.0, "tie_embeddings": true }, "train_args": { "steps": 1000, "batch_size": 8, "grad_accum": 4, "seq_len": 512, "lr": 0.0003, "warmup": 100, "save_every": 200, "log_every": 10, "corpus": "curated", "resume": true, "use_8bit_adam": true }, "total_params": 496986880, "steps": [ { "step": 720, "loss": 4.28348183631897, "ema_loss": 4.180650027796879, "lr": 0.00021277697160092742, "elapsed_s": 17.697318077087402, "steps_per_sec": 0.5650573695088259, "vram_gb": 15.430634496 }, { "step": 730, "loss": 4.118941903114319, "ema_loss": 4.135945131418912, "lr": 0.00021030580199599434, "elapsed_s": 35.07432723045349, "steps_per_sec": 0.5702176372077319, "vram_gb": 15.430634496 }, { "step": 740, "loss": 3.6725871562957764, "ema_loss": 4.0054830158378945, "lr": 0.00020781740384119815, "elapsed_s": 52.46801733970642, "steps_per_sec": 0.5717768942127871, "vram_gb": 15.430634496 }, { "step": 750, "loss": 3.6983203291893005, "ema_loss": 3.8964221631363376, "lr": 0.00020531272340588638, "elapsed_s": 69.84280610084534, "steps_per_sec": 0.5727146750410399, "vram_gb": 15.430634496 }, { "step": 760, "loss": 3.6233853101730347, "ema_loss": 3.7636052763477563, "lr": 0.00020279271315110968, "elapsed_s": 87.27109622955322, "steps_per_sec": 0.5729273741271987, "vram_gb": 15.428668416 }, { "step": 770, "loss": 3.3926801681518555, "ema_loss": 3.6054279373031792, "lr": 0.00020025833136742718, "elapsed_s": 104.76516604423523, "steps_per_sec": 0.5727094440404559, "vram_gb": 15.428668416 }, { "step": 780, "loss": 3.1609978079795837, "ema_loss": 3.4584633810890293, "lr": 0.00019771054181049524, "elapsed_s": 122.30735993385315, "steps_per_sec": 0.5723285993406916, "vram_gb": 15.428668416 }, { "step": 790, "loss": 3.1603373289108276, "ema_loss": 3.309057160849107, "lr": 0.00019515031333457712, "elapsed_s": 139.78667426109314, "steps_per_sec": 0.5723006175150589, "vram_gb": 15.428668416 }, { "step": 800, "loss": 2.7271658778190613, "ema_loss": 3.1769729487969567, "lr": 0.00019257861952411492, "elapsed_s": 200.5202157497406, "steps_per_sec": 0.44883255118937515, "vram_gb": 15.428668416 }, { "step": 810, "loss": 3.0822166204452515, "ema_loss": 3.088718658479337, "lr": 0.00018999643832350203, "elapsed_s": 217.9351396560669, "steps_per_sec": 0.4588521160828604, "vram_gb": 15.428668416 }, { "step": 820, "loss": 2.85100781917572, "ema_loss": 3.003455477485217, "lr": 0.00018740475166519754, "elapsed_s": 235.42823815345764, "steps_per_sec": 0.4672336711295415, "vram_gb": 15.428668416 }, { "step": 830, "loss": 2.7287580966949463, "ema_loss": 2.9638366754161387, "lr": 0.0001848045450963247, "elapsed_s": 252.95333313941956, "steps_per_sec": 0.4743958045963361, "vram_gb": 15.428668416 }, { "step": 840, "loss": 2.7534950971603394, "ema_loss": 2.85498482454621, "lr": 0.00018219680740389431, "elapsed_s": 270.4455711841583, "steps_per_sec": 0.48068821918875965, "vram_gb": 15.428668416 }, { "step": 850, "loss": 2.6869542002677917, "ema_loss": 2.7344452315263985, "lr": 0.00017958253023879622, "elapsed_s": 287.96280574798584, "steps_per_sec": 0.48617389886985163, "vram_gb": 15.428668416 }, { "step": 860, "loss": 2.446716547012329, "ema_loss": 2.6342544765284797, "lr": 0.00017696270773870167, "elapsed_s": 305.5090491771698, "steps_per_sec": 0.4909838199686599, "vram_gb": 15.428668416 }, { "step": 870, "loss": 2.5103383660316467, "ema_loss": 2.5549315583614782, "lr": 0.0001743383361500204, "elapsed_s": 323.075519323349, "steps_per_sec": 0.49524024703296865, "vram_gb": 15.428668416 }, { "step": 880, "loss": 2.2758803963661194, "ema_loss": 2.480248449238682, "lr": 0.00017171041344905526, "elapsed_s": 340.6705768108368, "steps_per_sec": 0.4990157987562144, "vram_gb": 15.428668416 }, { "step": 890, "loss": 2.280333399772644, "ema_loss": 2.428152495952108, "lr": 0.00016907993896249943, "elapsed_s": 358.2743320465088, "steps_per_sec": 0.5024083053112317, "vram_gb": 15.428668416 }, { "step": 900, "loss": 2.089028149843216, "ema_loss": 2.309313885635373, "lr": 0.00016644791298741992, "elapsed_s": 375.8712692260742, "steps_per_sec": 0.5054922138401625, "vram_gb": 15.428668416 }, { "step": 910, "loss": 2.392301082611084, "ema_loss": 2.231137390297292, "lr": 0.00016381533641087205, "elapsed_s": 393.4869828224182, "steps_per_sec": 0.5082760262243811, "vram_gb": 15.428668416 }, { "step": 920, "loss": 2.169742077589035, "ema_loss": 2.1655222685266406, "lr": 0.00016118321032928992, "elapsed_s": 411.1121575832367, "steps_per_sec": 0.5108095105591274, "vram_gb": 15.428668416 }, { "step": 930, "loss": 1.7299185693264008, "ema_loss": 2.0710785211805653, "lr": 0.00015855253566779667, "elapsed_s": 428.7321665287018, "steps_per_sec": 0.5131408771617605, "vram_gb": 15.428668416 }, { "step": 940, "loss": 2.0483551025390625, "ema_loss": 1.9999614091930757, "lr": 0.00015592431279958026, "elapsed_s": 446.3725197315216, "steps_per_sec": 0.5152646944716431, "vram_gb": 15.428668416 }, { "step": 950, "loss": 1.770757019519806, "ema_loss": 1.942065600070645, "lr": 0.00015329954116547926, "elapsed_s": 464.01725935935974, "steps_per_sec": 0.5172221402526133, "vram_gb": 15.428668416 }, { "step": 960, "loss": 1.6447540819644928, "ema_loss": 1.9088268255451808, "lr": 0.00015067921889392276, "elapsed_s": 481.6594491004944, "steps_per_sec": 0.5190389194416895, "vram_gb": 15.428668416 }, { "step": 970, "loss": 1.758211225271225, "ema_loss": 1.837405344969316, "lr": 0.00014806434242136985, "elapsed_s": 499.30819845199585, "steps_per_sec": 0.5207204704550765, "vram_gb": 15.428668416 }, { "step": 980, "loss": 1.5929311215877533, "ema_loss": 1.7585266107130897, "lr": 0.0001454559061133919, "elapsed_s": 516.9588961601257, "steps_per_sec": 0.522285237773273, "vram_gb": 15.428668416 }, { "step": 990, "loss": 1.3716070353984833, "ema_loss": 1.664784491957035, "lr": 0.00014285490188654308, "elapsed_s": 534.6144878864288, "steps_per_sec": 0.5237418856847029, "vram_gb": 15.428668416 }, { "step": 1000, "loss": 1.5660839676856995, "ema_loss": 1.6258334135726564, "lr": 0.00014026231883116147, "elapsed_s": 592.9495124816895, "steps_per_sec": 0.48908042572841365, "vram_gb": 15.428668416 }, { "step": 1010, "loss": 1.3805969655513763, "ema_loss": 1.5515371576395063, "lr": 0.00013767914283524546, "elapsed_s": 610.4368507862091, "steps_per_sec": 0.4914513263961972, "vram_gb": 15.428668416 }, { "step": 1020, "loss": 1.6006771624088287, "ema_loss": 1.4895273508195535, "lr": 0.0001351063562095476, "elapsed_s": 627.9561829566956, "steps_per_sec": 0.4936650174226852, "vram_gb": 15.428668416 }, { "step": 1030, "loss": 1.2836307883262634, "ema_loss": 1.4065686388898713, "lr": 0.00013254493731402865, "elapsed_s": 645.4572558403015, "steps_per_sec": 0.4957725660445192, "vram_gb": 15.428668416 }, { "step": 1040, "loss": 1.3605864644050598, "ema_loss": 1.3870377512296692, "lr": 0.0001299958601858145, "elapsed_s": 662.9888000488281, "steps_per_sec": 0.49774596490271933, "vram_gb": 15.428668416 }, { "step": 1050, "loss": 1.2070442289113998, "ema_loss": 1.3173434739374783, "lr": 0.00012746009416879654, "elapsed_s": 680.5442576408386, "steps_per_sec": 0.4996001306052267, "vram_gb": 15.428668416 }, { "step": 1060, "loss": 1.0203208923339844, "ema_loss": 1.2372483979334374, "lr": 0.0001249386035450169, "elapsed_s": 698.1243305206299, "steps_per_sec": 0.5013433634936998, "vram_gb": 15.428668416 }, { "step": 1070, "loss": 1.2167232036590576, "ema_loss": 1.1832412849618354, "lr": 0.00012243234716797824, "elapsed_s": 715.7207224369049, "steps_per_sec": 0.5029894883779004, "vram_gb": 15.428668416 }, { "step": 1080, "loss": 0.9296802133321762, "ema_loss": 1.13472966931652, "lr": 0.00011994227809801857, "elapsed_s": 733.3160738945007, "steps_per_sec": 0.504557329604138, "vram_gb": 15.428668416 }, { "step": 1090, "loss": 1.1032112538814545, "ema_loss": 1.0848564241444187, "lr": 0.00011746934323988837, "elapsed_s": 750.9138088226318, "steps_per_sec": 0.506050089284957, "vram_gb": 15.428668416 }, { "step": 1100, "loss": 0.7624894231557846, "ema_loss": 1.035134231257411, "lr": 0.0001150144829826688, "elapsed_s": 768.5393669605255, "steps_per_sec": 0.5074561132013314, "vram_gb": 15.428668416 }, { "step": 1110, "loss": 1.0394368767738342, "ema_loss": 0.9800935883989342, "lr": 0.00011257863084216771, "elapsed_s": 786.1641900539398, "steps_per_sec": 0.5087995676482738, "vram_gb": 15.428668416 }, { "step": 1120, "loss": 0.8956251740455627, "ema_loss": 0.9325636114854202, "lr": 0.0001101627131059291, "elapsed_s": 803.8073284626007, "steps_per_sec": 0.5100724831461603, "vram_gb": 15.428668416 }, { "step": 1130, "loss": 0.7532132491469383, "ema_loss": 0.8859757228483139, "lr": 0.00010776764848099159, "elapsed_s": 821.4637365341187, "steps_per_sec": 0.5112824599805761, "vram_gb": 15.428668416 }, { "step": 1140, "loss": 0.7787892669439316, "ema_loss": 0.858024045693639, "lr": 0.00010539434774452933, "elapsed_s": 839.1095054149628, "steps_per_sec": 0.5124480145024136, "vram_gb": 15.428668416 }, { "step": 1150, "loss": 0.7684804052114487, "ema_loss": 0.8112544161741339, "lr": 0.0001030437133975086, "elapsed_s": 856.7590296268463, "steps_per_sec": 0.5135633063495555, "vram_gb": 15.428668416 }, { "step": 1160, "loss": 0.6312309950590134, "ema_loss": 0.7749217716574573, "lr": 0.00010071663932149149, "elapsed_s": 874.4188885688782, "steps_per_sec": 0.5146274924784557, "vram_gb": 15.428668416 }, { "step": 1170, "loss": 0.6993876099586487, "ema_loss": 0.7453027083515941, "lr": 9.841401043871715e-05, "elapsed_s": 892.0863609313965, "steps_per_sec": 0.5156451439519039, "vram_gb": 15.428668416 }, { "step": 1180, "loss": 0.5929347425699234, "ema_loss": 0.7042416589248688, "lr": 9.613670237559053e-05, "elapsed_s": 909.741069316864, "steps_per_sec": 0.5166305181241613, "vram_gb": 15.428668416 }, { "step": 1190, "loss": 0.6512271761894226, "ema_loss": 0.6882631183908091, "lr": 9.388558112970555e-05, "elapsed_s": 927.3969824314117, "steps_per_sec": 0.5175777030690304, "vram_gb": 15.428668416 }, { "step": 1200, "loss": 0.6192804276943207, "ema_loss": 0.6477409202862138, "lr": 9.166150274053016e-05, "elapsed_s": 986.2773616313934, "steps_per_sec": 0.4968176489314273, "vram_gb": 15.428668416 }, { "step": 1210, "loss": 0.5546711385250092, "ema_loss": 0.6117017021277563, "lr": 8.946531296387777e-05, "elapsed_s": 1003.7613523006439, "steps_per_sec": 0.49812637122757175, "vram_gb": 15.428668416 }, { "step": 1220, "loss": 0.481452077627182, "ema_loss": 0.5780685955709823, "lr": 8.729784695029022e-05, "elapsed_s": 1021.3000912666321, "steps_per_sec": 0.4993635116271165, "vram_gb": 15.428668416 }, { "step": 1230, "loss": 0.45283564925193787, "ema_loss": 0.5569614956021417, "lr": 8.515992892745223e-05, "elapsed_s": 1038.7958240509033, "steps_per_sec": 0.5005796018434118, "vram_gb": 15.428668416 }, { "step": 1240, "loss": 0.47686392813920975, "ema_loss": 0.5290220321572775, "lr": 8.305237188676093e-05, "elapsed_s": 1056.3251626491547, "steps_per_sec": 0.5017394441980485, "vram_gb": 15.428668416 }, { "step": 1250, "loss": 0.3735135644674301, "ema_loss": 0.49133690958435916, "lr": 8.097597727416727e-05, "elapsed_s": 1073.8807201385498, "steps_per_sec": 0.5028491431807532, "vram_gb": 15.428668416 }, { "step": 1260, "loss": 0.45909684151411057, "ema_loss": 0.48467543677400826, "lr": 7.893153468540834e-05, "elapsed_s": 1091.462209224701, "steps_per_sec": 0.5039111710433675, "vram_gb": 15.428668416 }, { "step": 1270, "loss": 0.3922751322388649, "ema_loss": 0.45441256367994354, "lr": 7.691982156574558e-05, "elapsed_s": 1109.0576066970825, "steps_per_sec": 0.5049331942889357, "vram_gb": 15.428668416 }, { "step": 1280, "loss": 0.37874237447977066, "ema_loss": 0.4349958523394238, "lr": 7.494160291432407e-05, "elapsed_s": 1126.652606010437, "steps_per_sec": 0.5059234736237052, "vram_gb": 15.428668416 }, { "step": 1290, "loss": 0.35153336822986603, "ema_loss": 0.4098208888180574, "lr": 7.299763099326428e-05, "elapsed_s": 1144.2501864433289, "steps_per_sec": 0.5068821546822843, "vram_gb": 15.428668416 }, { "step": 1300, "loss": 0.4052524045109749, "ema_loss": 0.3939671492847197, "lr": 7.108864504159727e-05, "elapsed_s": 1161.8791389465332, "steps_per_sec": 0.5077980834865048, "vram_gb": 15.428668416 }, { "step": 1310, "loss": 0.32005714252591133, "ema_loss": 0.3771534058817519, "lr": 6.921537099415279e-05, "elapsed_s": 1179.5162498950958, "steps_per_sec": 0.5086831148391241, "vram_gb": 15.428668416 }, { "step": 1320, "loss": 0.2465466745197773, "ema_loss": 0.35445263362599905, "lr": 6.737852120550606e-05, "elapsed_s": 1197.1513810157776, "steps_per_sec": 0.5095429113421043, "vram_gb": 15.428668416 }, { "step": 1330, "loss": 0.3049493618309498, "ema_loss": 0.3454900266396032, "lr": 6.55787941790888e-05, "elapsed_s": 1214.7954704761505, "steps_per_sec": 0.510373980697331, "vram_gb": 15.428668416 }, { "step": 1340, "loss": 0.3119439110159874, "ema_loss": 0.3349345600271566, "lr": 6.381687430156783e-05, "elapsed_s": 1232.4373080730438, "steps_per_sec": 0.511182188232378, "vram_gb": 15.428668416 }, { "step": 1350, "loss": 0.2921401411294937, "ema_loss": 0.31941364171661696, "lr": 6.209343158259153e-05, "elapsed_s": 1250.084626674652, "steps_per_sec": 0.511965339260641, "vram_gb": 15.428668416 }, { "step": 1360, "loss": 0.24287717416882515, "ema_loss": 0.3099492300999808, "lr": 6.0409121400003627e-05, "elapsed_s": 1267.7480945587158, "steps_per_sec": 0.5127201553604033, "vram_gb": 15.428668416 }, { "step": 1370, "loss": 0.22612635791301727, "ema_loss": 0.2947590519243712, "lr": 5.876458425062101e-05, "elapsed_s": 1285.4061183929443, "steps_per_sec": 0.513456401487456, "vram_gb": 15.428668416 }, { "step": 1380, "loss": 0.22580422461032867, "ema_loss": 0.2781091778342552, "lr": 5.7160445506670415e-05, "elapsed_s": 1303.058067560196, "steps_per_sec": 0.514175090642343, "vram_gb": 15.428668416 }, { "step": 1390, "loss": 0.2514393739402294, "ema_loss": 0.26574903508551445, "lr": 5.559731517797666e-05, "elapsed_s": 1320.7140231132507, "steps_per_sec": 0.5148730066461105, "vram_gb": 15.428668416 }, { "step": 1400, "loss": 0.24877634644508362, "ema_loss": 0.2662288530131991, "lr": 5.4075787679992615e-05, "elapsed_s": 1380.0000896453857, "steps_per_sec": 0.49999996751978987, "vram_gb": 15.428668416 }, { "step": 1410, "loss": 0.29013000428676605, "ema_loss": 0.2518670179663263, "lr": 5.259644160775935e-05, "elapsed_s": 1397.4809682369232, "steps_per_sec": 0.5009012758743523, "vram_gb": 15.428668416 }, { "step": 1420, "loss": 0.21814599260687828, "ema_loss": 0.23880236520769688, "lr": 5.1159839515882365e-05, "elapsed_s": 1415.0194308757782, "steps_per_sec": 0.5017598942514659, "vram_gb": 15.428668416 }, { "step": 1430, "loss": 0.20425041019916534, "ema_loss": 0.2289297893300412, "lr": 4.9766527704607636e-05, "elapsed_s": 1432.520180940628, "steps_per_sec": 0.5026107203091759, "vram_gb": 15.428668416 }, { "step": 1440, "loss": 0.21563759073615074, "ema_loss": 0.22410501963253138, "lr": 4.8417036012078574e-05, "elapsed_s": 1450.0505871772766, "steps_per_sec": 0.5034307123181445, "vram_gb": 15.428668416 }, { "step": 1450, "loss": 0.21186616644263268, "ema_loss": 0.2213783268398314, "lr": 4.711187761285319e-05, "elapsed_s": 1467.6139550209045, "steps_per_sec": 0.5042197898625592, "vram_gb": 15.428668416 }, { "step": 1460, "loss": 0.17390306666493416, "ema_loss": 0.20964586315877706, "lr": 4.5851548822757895e-05, "elapsed_s": 1485.1958084106445, "steps_per_sec": 0.5049839191255185, "vram_gb": 15.428668416 }, { "step": 1470, "loss": 0.19438401982188225, "ema_loss": 0.20803681597852408, "lr": 4.463652891015232e-05, "elapsed_s": 1502.7898688316345, "steps_per_sec": 0.5057260604177967, "vram_gb": 15.428668416 }, { "step": 1480, "loss": 0.18505235388875008, "ema_loss": 0.20459605664240096, "lr": 4.346727991367676e-05, "elapsed_s": 1520.390971660614, "steps_per_sec": 0.5064486795518025, "vram_gb": 15.428668416 }, { "step": 1490, "loss": 0.1997264064848423, "ema_loss": 0.19819135992439754, "lr": 4.234424646655138e-05, "elapsed_s": 1538.0068547725677, "steps_per_sec": 0.5071498853074632, "vram_gb": 15.428668416 }, { "step": 1500, "loss": 0.16755592823028564, "ema_loss": 0.1918936216767847, "lr": 4.1267855627494735e-05, "elapsed_s": 1555.638013601303, "steps_per_sec": 0.5078302234149894, "vram_gb": 15.428668416 }, { "step": 1510, "loss": 0.17541416734457016, "ema_loss": 0.18443936061329902, "lr": 4.023851671832496e-05, "elapsed_s": 1573.2746274471283, "steps_per_sec": 0.5084935497231776, "vram_gb": 15.428668416 }, { "step": 1520, "loss": 0.14539901167154312, "ema_loss": 0.17578462170763828, "lr": 3.925662116830581e-05, "elapsed_s": 1590.9256908893585, "steps_per_sec": 0.5091375446625632, "vram_gb": 15.428668416 }, { "step": 1530, "loss": 0.1841065064072609, "ema_loss": 0.17096551335138493, "lr": 3.8322542365297113e-05, "elapsed_s": 1608.5824146270752, "steps_per_sec": 0.5097656125938094, "vram_gb": 15.428668416 }, { "step": 1540, "loss": 0.15982552990317345, "ema_loss": 0.16729554632520202, "lr": 3.743663551376553e-05, "elapsed_s": 1626.2370433807373, "steps_per_sec": 0.5103806996515938, "vram_gb": 15.428668416 }, { "step": 1550, "loss": 0.17187346518039703, "ema_loss": 0.1643358000447934, "lr": 3.6599237499710016e-05, "elapsed_s": 1643.8927867412567, "steps_per_sec": 0.5109822287529833, "vram_gb": 15.428668416 }, { "step": 1560, "loss": 0.1427095215767622, "ema_loss": 0.15744941248457628, "lr": 3.581066676255348e-05, "elapsed_s": 1661.5445902347565, "steps_per_sec": 0.5115721871056769, "vram_gb": 15.428668416 }, { "step": 1570, "loss": 0.12492887675762177, "ema_loss": 0.15700561553855466, "lr": 3.5071223174048926e-05, "elapsed_s": 1679.1984930038452, "steps_per_sec": 0.5121491018382129, "vram_gb": 15.428668416 }, { "step": 1580, "loss": 0.14507505670189857, "ema_loss": 0.15353575625923596, "lr": 3.438118792424645e-05, "elapsed_s": 1696.8557302951813, "steps_per_sec": 0.5127130046870023, "vram_gb": 15.428668416 }, { "step": 1590, "loss": 0.14994086883962154, "ema_loss": 0.15015913800263153, "lr": 3.374082341456457e-05, "elapsed_s": 1714.52734375, "steps_per_sec": 0.5132609889296202, "vram_gb": 15.428668416 }, { "step": 1600, "loss": 0.14491559006273746, "ema_loss": 0.1493680333619406, "lr": 3.315037315800602e-05, "elapsed_s": 1776.030238866806, "steps_per_sec": 0.501117593903054, "vram_gb": 15.428668416 }, { "step": 1610, "loss": 0.13732506521046162, "ema_loss": 0.14832111416366284, "lr": 3.2610061686556506e-05, "elapsed_s": 1793.507203578949, "steps_per_sec": 0.5018100837309419, "vram_gb": 15.428668416 }, { "step": 1620, "loss": 0.1437094621360302, "ema_loss": 0.14366319953712975, "lr": 3.2120094465801376e-05, "elapsed_s": 1811.0531499385834, "steps_per_sec": 0.5024700683306065, "vram_gb": 15.428668416 }, { "step": 1630, "loss": 0.1201076190918684, "ema_loss": 0.1377247703954839, "lr": 3.168065781679277e-05, "elapsed_s": 1828.5478429794312, "steps_per_sec": 0.503131489576425, "vram_gb": 15.428668416 }, { "step": 1640, "loss": 0.13061938621103764, "ema_loss": 0.13602811819866328, "lr": 3.129191884519671e-05, "elapsed_s": 1846.0855515003204, "steps_per_sec": 0.5037686358815742, "vram_gb": 15.428668416 }, { "step": 1650, "loss": 0.12817695923149586, "ema_loss": 0.1344849350545901, "lr": 3.095402537774757e-05, "elapsed_s": 1863.6434466838837, "steps_per_sec": 0.5043883268940796, "vram_gb": 15.428668416 }, { "step": 1660, "loss": 0.13230768591165543, "ema_loss": 0.13204115287728813, "lr": 3.066710590603351e-05, "elapsed_s": 1881.2273790836334, "steps_per_sec": 0.5049894609033149, "vram_gb": 15.428668416 }, { "step": 1670, "loss": 0.11046527698636055, "ema_loss": 0.1296894053225092, "lr": 3.0431269537634746e-05, "elapsed_s": 1898.828557729721, "steps_per_sec": 0.50557486935408, "vram_gb": 15.428668416 }, { "step": 1680, "loss": 0.13638370111584663, "ema_loss": 0.12728797289501487, "lr": 3.0246605954632904e-05, "elapsed_s": 1916.423189163208, "steps_per_sec": 0.5061512537966854, "vram_gb": 15.428668416 }, { "step": 1690, "loss": 0.12420566566288471, "ema_loss": 0.12502314340227097, "lr": 3.0113185379507383e-05, "elapsed_s": 1934.0418214797974, "steps_per_sec": 0.506710862772435, "vram_gb": 15.428668416 }, { "step": 1700, "loss": 0.10359455086290836, "ema_loss": 0.12350706497894226, "lr": 3.0031058548431695e-05, "elapsed_s": 1951.6798777580261, "steps_per_sec": 0.5072553195236368, "vram_gb": 15.428668416 }, { "step": 1710, "loss": 0.11198867857456207, "ema_loss": 0.12169409054894889, "lr": 3.0000256691979898e-05, "elapsed_s": 1969.3243565559387, "steps_per_sec": 0.5077883674525075, "vram_gb": 15.428668416 } ] }