题目: ALIGNING LARGE MULTIMODAL MODELS WITH FACTUALLY AUGMENTED RLHF
机构:UC伯克利
论文: https://arxiv.org/pdf/2309.14525.pdf
代码:https://llava-rlhf.github.io/
任务:
特点:
方法:
前置相关工作:
题目:RLHF-V: Towards Trustworthy MLLMs via Behavior Alignment from Fine-grained Correctional Human Feedback
机构:清华,新加坡国立
论文?