{"slug": "in-place-instruction-following-in-diffusion-language-models", "title": "In-Place Instruction Following in Diffusion Language Models", "summary": "A new arXiv paper (2609.07160v1) formalizes In-place Instruction Following (IIF) for diffusion large language models and introduces GRAFT, a post-training framework that raised the average IIF score from 57.75 to 73.10 (+15.35 points) across four representative dLLMs. The authors also built IIF-Bench, a hierarchical benchmark covering literal, style, and discourse-function constraints with a rubric-based local-global evaluation protocol, and an inference-time attention-bias probe indicating vanilla dLLMs often under-prioritize constraint spans during denoising. GRAFT combines constraint-aware supervised fine-tuning with preference optimization, delivering absolute gains of 15.91 points on literal constraints and 15.57 points on discourse-function constraints while preserving general generation ability.", "body_md": "arXiv:2609.07160v1 Announce Type: cross \nAbstract: 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.", "url": "https://wpnews.pro/news/in-place-instruction-following-in-diffusion-language-models", "canonical_source": "https://www.machinebrief.com/news/in-place-instruction-following-in-diffusion-language-models-3zn2", "published_at": "2026-09-10 04:00:00+00:00", "updated_at": "2026-09-10 06:21:46.219748+00:00", "lang": "en", "topics": ["large-language-models", "ai-research", "natural-language-processing", "generative-ai"], "entities": ["arXiv", "IIF-Bench", "GRAFT", "In-place Instruction Following", "In-place Prompting"], "alternates": {"html": "https://wpnews.pro/news/in-place-instruction-following-in-diffusion-language-models", "markdown": "https://wpnews.pro/news/in-place-instruction-following-in-diffusion-language-models.md", "text": "https://wpnews.pro/news/in-place-instruction-following-in-diffusion-language-models.txt", "jsonld": "https://wpnews.pro/news/in-place-instruction-following-in-diffusion-language-models.jsonld"}}