ResearchBenchmarksLanguage Modeling on OWT L=10240 (test)Follow23.4Gen. PPLEso-LM23.15624.80326.4528.097Jun 2, 2025Evaluation ResultsMethodMethodLinksGen. PPLEntropyTime (seconds)Eso-LMalpha_0_train=0.125, a...alpha_0_train=0.125, alpha_0_eval=0.125, Sequence length (L)=10240, NFEs=102402025.0623.46.3116.4BD3-LMsL'=4, Sequence length...L'=4, Sequence length (L)=10240, NFEs=102402025.0629.56.5588.6