作者位次 #6/9ACL 2026Poster· Natural Language Processing
Saber: Efficient Sampling with Adaptive Acceleration and Backtracking Enhanced Remasking for Diffusion Language Model in Code Generation
Yihong Dong · Zhaoyu Ma · Xue Jiang · Zhiyuan Fan · Jiaru Qian · Yongmin Li · Jianha Xiao · Zhi Jin · Ge Li
摘要
Diffusion language models (DLMs) are emerging as a compelling alternative to the dominant autoregressive paradigm, offering inherent advantages in parallel generation and bidirectional context modeling. However, for the tasks with strict structural constraints such as code generation, DLMs face a...