Multimodal in-context learning (ICL) remains underexplored despite the profound potential it could have in complex application domains such as medicine. Clinicians routinely face a long tail of tasks which they need to learn to solve from few examples, such as considering few relevant previous cases or few differential diagnoses. While MLLMs… See the full description on the dataset page:
https://huggingface.co/datasets/smmile/SMMILE-plusplus.
