DA-DPO: Cost-efficient Difficulty-aware Preference Optimization for Reducing MLLM Hallucinations

Longtian Qiu, Shan Ning, Chuyu Zhang, Jiaxuan Sun, Xuming He 0001. DA-DPO: Cost-efficient Difficulty-aware Preference Optimization for Reducing MLLM Hallucinations. Trans. Mach. Learn. Res., 2025, 2025. [doi]

Abstract

Abstract is missing.