arXiv - CSCL: "RLHF-V: Towards Trustworthy MLLMs via Behavior Al…" - Qoto Mastodon

arXiv - CSCL @arxiv_cscl@qoto.org

RLHF-V: Towards Trustworthy MLLMs via Behavior Alignment from Fine-grained Correctional Human Feedback. (arXiv:2312.00849v1 [cs.CL])

http://arxiv.org/abs/2312.00849 #arXiv #NLProc

Dec 05, 2023, 03:19 · · arxiv-cscl · · ·

Sign in to participate in the conversation