Corentin Dancette

Beyond task performance: evaluating and reducing the flaws of large multimodal models with in-context-learning