2 papers across 1 session
We propose MMPB, the first benchmark for evaluating personalization in large vision–language models.