Mitigating Entangled Steering in Large Vision-Language Models for Hallucination Reduction

Ähnliche Einträge

• Dynamic Multimodal Activation Steering for Hallucination Mitigation in Large Vision-Language Models
  von: Yin, Jianghao, et al.
  Veröffentlicht: (2026)
• Adaptive Residual-Update Steering for Low-Overhead Hallucination Mitigation in Large Vision Language Models
  von: Zou, Zhengtao, et al.
  Veröffentlicht: (2025)
• Prefill-Time Intervention for Mitigating Hallucination in Large Vision-Language Models
  von: Zhang, Chengsheng, et al.
  Veröffentlicht: (2026)
• INTER: Mitigating Hallucination in Large Vision-Language Models by Interaction Guidance Sampling
  von: Dong, Xin, et al.
  Veröffentlicht: (2025)
• Vision-Language Introspection: Mitigating Overconfident Hallucinations in MLLMs via Interpretable Bi-Causal Steering
  von: Liu, Shuliang, et al.
  Veröffentlicht: (2026)