Plan-and-Patch: Diffusion Language Models for Agentic Planning
We introduce Plan-and-Patch, a plan-and-act framework in which a diffusion language model (dLLM) generates a structured, program-like plan through parallel unmasking and repairs it by filling in selected regions while keeping the surrounding steps fixed.
Key points
- Planning is increasingly important for long-horizon agents, where successful execution requires coordinating subgoals, tool use, and intermediate outcomes over many steps.
- We compare DreamReasoner-8B and Qwen3-8B as diffusion and autoregressive (AR) planners.
- After task-specific training on agentic benchmarks, ALFWorld and TextCraft, the planners achieve similar observed success in plan generation, while diffusion reduces mean plan-generation latency by 39-46% relative to AR.
- Our results show that Plan-and-Patch provides a framework for faster plan generation and effective plan repair in long-horizon agents.
Sources (1)
- [1]Plan-and-Patch: Diffusion Language Models for Agentic PlanningarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 06:45 PM
We introduce Plan-and-Patch, a plan-and-act framework in which a diffusion language model (dLLM) generates a structured, program-like plan through parallel unmasking and repairs it by filling in selected regions while keeping the surrounding steps fixed.
Planning is increasingly important for long-horizon agents, where successful execution requires coordinating subgoals, tool use, and intermediate outcomes over many steps.
Extractive summary: sentences quoted from the sources.