DLoop: Looped Speculative Decoding
Speculative decoding accelerates autoregressive generation in large language models. In each drafting stage, a lightweight draft model proposes tokens that the target model subsequently verifies.
By Geonmo Gu, Byeongho Heo, HeeJae Jun, Yoohoon Kang, Sangmin Lee, Sangdoo Yun, Dongyoon HanarXiv - cs.CL (NLP)
Read the full story
Originally published by arXiv - cs.CL (NLP) on Oct 7, 2026.
Comments
Sign in to join the discussion.