TY  - RPRT
TI  - DA-DPO: Cost-efficient Difficulty-aware Preference Optimization for Reducing MLLM Hallucinations
AU  - Longtian Qiu
AU  - Shan Ning
AU  - Chuyu Zhang
AU  - Jiaxuan Sun
AU  - Xuming He
PY  - 2026
UR  - https://arxiv.org/abs/2601.00623
ID  - 2601.00623
ER  -