English(EN) HarnessPro: Recursive Harness Self-Improvement with Pareto Optimization for Performance-Cost Trade-offs

HarnessPro 提升 AI 代理性能并降低成本

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-21 04:00

研究人员开发了 HarnessPro，这是一种通过递归增强性能同时降低执行成本来改进 AI 代理 Harness 的新方法。这种无需标签的方法利用 LLM 生成的测试和专用代理来诊断和规划更新，提供针对不同预算限制定制的“lite”、“medium”和“heavy”Harness 配置。在 SWE-Bench Pro 等基准测试上的实验表明，HarnessPro 使用 GPT-5.5 结合 Codex 等模型，可以在通过率显著提高的同时大幅降低成本。 AI

影响该方法可能为各种任务带来更高效、更具成本效益的 AI 代理。

排序理由该集群包含一篇详细介绍 AI 代理开发新方法的论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.CL 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 0 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该集群包含一篇详细介绍 AI 代理开发新方法的论文。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, product; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: 5 days old; Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.CL TIER_1 English(EN) · Zhuoshi Pan, ZhiYuan Feng, Yukuo Cen, Honglin Lin, Junru Lu, Zhongxia Chen, Shuo Yang, H. Vicky Zhao, Jie Tang · 2026-10-21 04:00

HarnessPro：通过帕累托优化实现递归式安全带自我改进，以平衡性能与成本

arXiv:2610.26770v2 Announce Type: replace-cross Abstract: Harness development plays a crucial role in building capable AI agents. Recent approaches improve harnesses through recursive self-improvement (RSI), but their performance gains often come with higher execution costs. We i…

本文如何被排名

报道来源 [1]

HarnessPro：通过帕累托优化实现递归式安全带自我改进，以平衡性能与成本

相关实体

相关话题