MiniGPT-Med

A radiology generalist vision-language model from the MiniGPT lineage that unifies report generation, visual question answering, and disease identification in one model. The 140★ release is a standard citation in medical-VLM surveys.

Adapted from the MiniGPT-4 recipe over pretrained vision and language bases rather than pretrained from scratch, so its scale is inherited from those components. Part of the Mohamed Elhoseiny / Vision-CAIR line.