TY  - RPRT
TI  - Reducing Hallucination in Multimodal Large Language Models through Hard Grounding Preference Supervision
AU  - Qinwu Xu
PY  - 2026
UR  - https://arxiv.org/abs/2605.16411
ID  - 2605.16411
ER  -