AdaptLLM/biomed-VQA-benchmark

该数据集主要用于生物医学领域的视觉问答任务，包含多个配置（如SLAKE、VQA_RAD、PathVQA、PMC-VQA）。数据集通过图像-标题对生成多样化的视觉指令任务，用于评估多模态大语言模型（MLLMs）在特定领域的表现。数据集的使用场景包括数据合成、训练管道和任务评估，特别是在生物医学和食品领域的应用。

This project contains multiple datasets for evaluating multimodal large language models on biomedical visual instruction tasks, including SLAKE, VQA_RAD, PathVQA, and PMC-VQA. These datasets are used for testing and evaluating model performance on various domain-specific tasks, supporting domain adaptation through post-training methods.