arXiv 2610.27103 · cs.AI · 2026-10-19
Omni-Woodpecker: Hallucination Correction via Evidence-Prior Rebalancing for Omni LLMs
Zijian Liu, Chaoyou Fu, Yan Yang, Caifeng Shan, Ran He
Nanjing University
一句话要点
Omni-Woodpecker通过证据优先重平衡机制，纠正多模态大模型中的幻觉问题，在多个基准上显著提升性能且具备跨架构泛化能力。
Omni-Woodpecker corrects hallucinations via evidence-prior rebalancing, improving performance across benchmarks with strong cross-architecture generalization.
要点
- 多模态大模型在联合音频视觉推理中仍存在显著的跨模态幻觉问题，现有方法多在输出层面处理，忽视了证据与先验之间的纠缠关系。
- Omni-Woodpecker提出一种无需训练的框架，通过识别模态间冲突、利用正交分解分离证据与先验，并动态调整融合表示以抑制误导性先验并恢复被忽略的证据。
- 在Qwen2.5-Omni-7B模型上，OWP在CMM和AVHBench上分别提升12.9和8.6个点，在四个通用推理基准上持续获得0.67至2.25个点的增益。
- 该方法在VideoLLaMA2-AV模型上也表现出良好泛化能力，分别在CMM和AVHBench上提升4.6和3.9个点，验证了其跨架构有效性。
English overview
Omni-Woodpecker addresses cross-modal hallucinations in Omni LLMs by rebalancing fused representations based on evidence-prior conflicts. It identifies modality-specific conflicts, decomposes hidden states into grounding evidence and competing priors, and adaptively suppresses misleading priors while restoring evidence before decoding. The method achieves significant performance gains across multiple benchmarks and generalizes well to different architectures without requiring retraining.
只根据 arXiv 官方摘要，由开源模型整理，经数字/专名自动核对（必须在摘要中出现），以论文原文为准。
阅读 arXiv 原文 PDF Hugging Face 讨论 代码仓库
主题与学科
同主题：多模态
- MJev-Spatial: System One Spatial Intelligence via Native Attention Matching ▲556
- LoopVL: Recurrent Visual Intelligence ▲417
- VideoJev: System One Model for Video Understanding and Interaction ▲375
- S-Space: Exploring Spatial Workspace in Multimodal Language Models ▲270
- UniEvo-VL: An On-policy Self-Distillation Training Recipe for Multimodal Model Self-improvement ▲266
同主题：安全与可解释
- Post-Training Leaves Behavioral Shadows on Unrelated Decisions 2026-09-24 · ▲273
- Right Screen, Wrong Transition: World Models as Verifiers for GUI Agents 2026-10-08 · ▲188
- Behavioral Compositionality of Sparse Autoencoder Representations 2026-10-15 · ▲108
- DecepEval: A Benchmark for Evaluating Deception in LLM Agents 2026-10-06 · ▲85
- Last Secure Code Benchmark: Is the Code You Vibe Secure? 2026-10-10 · ▲76
同机构：Nanjing University
数据说明
元数据来自 arXiv API（已核实）；点赞来自 Hugging Face 每日论文；引用来自 Semantic Scholar。最近刷新 2026-10-19。
要点由本站根据摘要自写，不转载摘要原文，请阅读原文。