English(EN) SHAM-Bench: A Span-level Hallucination Benchmark Across Multiple Tasks

新的SHAM-Bench基准评估跨任务的LLM幻觉检测能力

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

研究人员推出SHAM-Bench，一个旨在评估大型语言模型在多个自然语言生成任务中跨度级幻觉检测能力的新基准。该基准在通用模式和评估协议下整合了来自现有数据集的6,461个示例。使用SHAM-Bench进行的初步评估显示，虽然领域内微调的检测器表现良好，但在分布变化下性能会下降，而具有推理能力的提示式方法在不同任务中表现出更强的鲁棒性。 AI

影响该基准旨在提高LLM输出中事实性不准确的检测和缓解能力，从而实现更可靠的AI系统。

排序理由该集群描述了一篇介绍用于评估AI模型基准数据集的新学术论文。[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; 该集群描述了一篇介绍用于评估AI模型基准数据集的新学术论文。[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, 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: 13 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) · Sicong Huang, Ian Lane · 2026-10-21 04:00

SHAM-Bench：跨任务的跨度级幻觉基准

arXiv:2610.27370v1 Announce Type: new Abstract: Span-level hallucination detection pinpoints unsupported content in LLM outputs, enabling actionable feedback and targeted mitigation. Existing evaluations for span-level hallucination detection methods are fragmented across tasks, …

本文如何被排名

报道来源 [1]

SHAM-Bench：跨任务的跨度级幻觉基准

相关实体

相关话题