Eso-LMs combine autoregressive and masked diffusion models, improving perplexity and inference efficiency with KV caching, achieving state-of-the-art performance and significantly faster inference rates. Code and checkpoints available online.

https://arxiv.org/abs//2506.01928

YouTube: https://www.youtube.com/@ArxivPapers

TikTok: https://www.tiktok.com/@arxiv_papers

Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016

Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers