遇见数据集

MiaoMiaoYang/CXR-MAX

收藏
Hugging Face2025-09-29 更新2025-10-25 收录
官方服务:

资源简介:

CXR-MAX(多教师对X射线的对齐)数据集是一个包含来自七个广泛使用公共多模态大型语言模型输出的MIMIC-CXR数据集的扩展。它提供了170,982个推理轨迹的知识蒸馏实例,涵盖了14种胸部病理学,为临床胸部X射线解释的知识蒸馏建立了首个大规模基准。

CXR-MAX (Multi-teachers Alignment for X-rays) Dataset is an extension of the MIMIC-CXR dataset incorporating outputs from seven widely used public multi-modal large language models. It provides 170,982 distillation instances of reasoning trajectories covering 14 thoracic pathologies, establishing the first large-scale benchmark for knowledge distillation in clinical chest X-ray interpretation.

提供机构:
MiaoMiaoYang
二维码
社区交流群
二维码
科研交流群
商业服务