共计 381 篇文章
2026
In-Place Test-Time Training
Beyond the Global Scores： Fine-Grained Token Grounding as a Robust Detector of LVLM Hallucinations
EMNLP 2025关于幻觉的论文
SAVE： Sparse Autoencoder-Driven Visual Information Enhancement for Mitigating Object Hallucination
Neural Uncertainty Principle： A Unified View of Adversarial Fragility and LLM Hallucination
DynHD： Hallucination Detection for Diffusion Large Language Models via Denoising Dynamics Deviation Learnig
Attention MoA
Locate-then-Sparsify： Attribution Guided Sparse Strategy for Visual Hallucination Mitigation
Fighting Hallucinations with Counterfactuals： Diffusion-Guided Perturbations for LVLM Hallucination Suppression
Efficient Refusal Ablation in LLM through Optimal Transport