[论文] AutoDesign: Meta-Harness Optimization for Long-Horizon Agentic Design

论文概要

研究领域: NLP
作者: Yaxin Luo, Haobin Jiang, Jialv Zou, Xu Huang, Wenhao Yan, Haodong Li, Zhengrong Yue, Jing Li, Xiaofu Chen, Xiaohan Zhao, Jiacheng Liu, Jiacheng Cui, Zhiqiang Shen, Xiaotong Li
发布时间: 2026-08-13
arXiv: 2608.13560

中文摘要

将多模态源转换为紧凑且结构化的媒体输出，从根本上可以被概念化为一个以模型Harness系统为核心的长程智能体过程。理想的Harness系统应当与人类设计先验对齐，并通过经验探索积累可复用的经验以实现递归式自我改进，但现有范式仍然静态化，无法达到这一能力。本文提出AutoDesign框架，通过元Harness优化器引导代码智能体基于 rollout 反馈递归改进Harness。为实例化和评估该框架，我们聚焦学术论文转海报生成任务，引入PosterBench基准，包含涵盖五个学科的100篇论文主赛道和一个共享的10篇论文子集用于对照评估。在PosterBench主赛道上，AutoDesign获得最高分78.32，超越闭源商业系统Claude Design达7.45分。在七种对照代码智能体-模型配置中，集成学习到的DesignHarness持续提升性能，平均PosterBench分数从54.99提升至67.39（+12.4%）。在完全自主的长程循环中，它在40分钟内执行253次工具调用和11轮编辑，成本不到3美元，达到了人类评估中的平均会议海报质量。系统盲测人类研究进一步证明，AutoDesign在所有评估系统中获得最高的人类偏好度。

原文摘要

Transforming multimodal sources into condensed and structured media outputs can be fundamentally conceptualized as a long-horizon agentic process centered on a model-harness system. While an ideal harness system should align with human design priors and accumulate reusable experience through empirical exploration to drive recursive self-improvement, existing paradigms remain static and fall short of this capability. In this paper, we present AutoDesign, a framework that aligns with human design priors, where a meta-harness optimizer guides a code agent to recursively improve harness based on rollout feedback. To instantiate and evaluate this framework, we focus on the academic paper-to-poster generation task and introduce PosterBench, comprising a 100-paper Main Track spanning five disciplines and PosterBench-mini, a shared 10-paper subset for controlled evaluation. On the PosterBench Main Track, AutoDesign achieves the highest score of 78.32, surpassing the closed-source commercial system Claude Design by 7.45 points. Across seven controlled code-agent-model configurations, integrating the learned DesignHarness consistently improves performance, increasing the average PosterBench Score from 54.99 to 67.39 (+12.4%). In a fully autonomous long-horizon loop, it executes 253 tool calls and 11 editing turns within 40 minutes for under $3, reaching average conference-poster quality in human evaluation. A system-blind human study further demonstrates that AutoDesign achieves the highest human preference among evaluated systems.

自动采集于 2026-08-15

#论文 #arXiv #NLP #小凯