English(EN) What Should We Remove? Evaluating the Reliability and Costs of RLVR Data Decontamination

新研究质疑 RLVR 数据去污方法的有效性

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-16 04:00

一篇新发布的 arXiv 论文评估了用于可验证奖励强化学习 (RLVR) 数学语料库的数据去污方法的有效性和成本。该研究使用精确匹配、词汇匹配和语义匹配，针对六个基准审计了十六个语料库，并评估了八种去污策略。结果表明，当前的语义匹配策略经常导致数据被错误移除，其中一种测试策略实现了 98.1% 的已解决精度和 98.2% 的候选帧召回率，显示出比其他方法更优的平衡。 AI

影响强调了人工智能训练中数据清理的潜在问题，影响了基准测试的可靠性和模型性能。

排序理由该集群包含一篇研究论文，详细介绍了对 RLVR 数学语料库的数据去污方法的评估。[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: 17 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该集群包含一篇研究论文，详细介绍了对 RLVR 数学语料库的数据去污方法的评估。[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, safety; 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.CL TIER_1 English(EN) · Esther Xin · 2026-10-16 04:00

我们应该移除什么？评估 RLVR 数据去污的可靠性和成本

arXiv:2610.21229v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards (RLVR) math corpora are routinely screened for benchmark overlap, but the reliability and data-loss costs of cleaning policies remain underexamined. We audit sixteen open RLVR math corp…

本文如何被排名

报道来源 [1]

我们应该移除什么？评估 RLVR 数据去污的可靠性和成本

相关实体

相关话题