PaperScope
LIVE · 2026-09-09 05:40 UTC

In-Place Instruction Following in Diffusion Language Models

Zheng Nie, Zherui Li, Jiaming Zhang, Kun Wang, Zhenhong Zhou, Yufei Guo

Latestcs.CLcs.LGcs.AIcs.CV
arXiv ID
2609.07160 v1
Category
Submitted
2026-09-07

Abstract

Diffusion Large Language Models (dLLMs) generate text via bidirectional iterative denoising, naturally supporting user-specified constraints anchored at arbitrary output positions, a paradigm known as In-place Prompting (IPP). We formalize this as the In-place Instruction Following (IIF) task and construct IIF-Bench, a hierarchical benchmark spanning literal, style, and discourse-function constraints, paired with a rubric-based local-global evaluation protocol. An inference-time attention-bias probe suggests that vanilla dLLMs often under-prioritize constraint spans during denoising. We then propose GRAFT, an IPP-oriented post-training framework combining constraint-aware SFT and preference optimization. On four representative dLLMs, GRAFT raises the average IIF score from 57.75 to 73.10 (+15.35 points), with absolute gains of 15.91 and 15.57 points on literal and discourse-function constraints, while preserving general generation ability.

arXiv abs page · PDF · same-day batch