LLaVA-Medical-8B CLIP-ViT (Stage 2)

A biomedical vision-language model that combines a CLIP ViT vision encoder with an 8B-parameter language model to answer questions about medical images and support multimodal medical reasoning. This checkpoint represents the second-stage instruction-tuned variant fine-tuned on medical data, following the LLaVA-Med two-stage training paradigm. Source: supabase.