RLHF-V: Towards Trustworthy MLLMs via Behavior Alignment from Fine-grained Correctional Human Feedback
Open paper
Benchmark
Model
Rank
Results
image-captioning-on-object-halbench
RLHF-V
#1
chair_i: 7.5
chair_s: 12.2
Rank counts only results with a code link.