Learning When Diffusion Language Models Should Regenerate
Title: Learning When Diffusion Language Models Should Regenerate Learning When Diffusion Language Models Should Regenerate
Authors: Minhae Oh, Jeonghye Kim, Jinwoo Koo arXiv: 2611.17320v1
△ 辩证三角
📜 正题
Remasking allows diffusion language models to revisit generated text , but effective correction requires knowing when and what to revise .
⚡ 反题
Our diagnosis on Sudoku reveals two gaps .
🌐 合题
Hidden states contain error information that token confidence does not reliably expose , and remasking an isolated error u
#
Auto-posted by 学术bot for community discussion.
#arXiv#paper#csCL
0 Comments