English(EN) Prompt Interference in LLM Post-Training: Why Pass@k Optimization Can Degrade Pass@1

LLM 研究揭示提示干扰会降低单次运行性能

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-15 04:00

一篇新的研究论文探讨了大型语言模型中的“提示干扰”现象，即针对 pass@k 指标的优化可能会无意中降低 pass@1 指标的性能。这是因为共享的模型参数连接着不同的提示，导致一些提示的改进会对其他提示产生负面影响。该研究为这种权衡提供了理论解释，表明 pass@k 优化可以通过隐式地重新加权成功率较低的提示，从而将模型的更新方向从 pass@1 转移开。 AI

影响强调了 LLM 评估中的一个关键权衡，可能会影响模型在具有延迟限制的实际应用中的微调方式。

排序理由关于 LLM 性能指标和优化权衡的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 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: 16 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 关于 LLM 性能指标和优化权衡的学术论文。[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, model release; 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: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Anas Barakat, Souradip Chakraborty, Khushbu Pahwa, Amrit Singh Bedi · 2026-10-15 04:00

LLM训练后提示干扰：为何Pass@k优化会降低Pass@1

arXiv:2602.21189v3 Announce Type: replace-cross Abstract: Pass@$k$ is a widely used performance metric for verifiable large language model tasks, including mathematical reasoning, code generation, and short-answer reasoning. It defines success if any of $k$ independently sampled …

本文如何被排名

报道来源 [1]

LLM训练后提示干扰：为何Pass@k优化会降低Pass@1

相关实体

相关话题