RubiCap: Rubric-Guided Reinforcement Learning for Dense Image Captioning
Open paper
Benchmark
Model
Rank
Results
multimodal-reasoning-on-mmmu
RubiCap-7B (PixMoCap)
–
Accuracy: 37.0
visual-question-answering-vqa-on-mmbench-en
RubiCap-7B (PixMoCap)
–
Accuracy: 67.61
Rank counts only results with a code link.