yyaxx/SlideChat

该数据集提供了用于训练和评估SlideChat模型的资源，SlideChat是一个用于全切片病理图像理解的多模态大语言模型。数据集包含指令跟随的训练数据和VQA/Caption评估基准，覆盖多个病理队列和任务。具体内容包括：训练指令数据，分为第一阶段使用的Slide-level caption instruction data（SlideInstruct_train_stage1_caption.json）和第二阶段使用的Slide-level VQA instruction data（SlideInstruct_train_stage2_vqa.json）；评估基准（SlideBench），包括用于TCGA、CPTAC和Histai队列的Caption/Report评估分割（SlideBench-Caption-*.csv），以及用于TCGA、BCNB、CPTAC和Histai队列的Closed VQA评估分割（SlideBench-VQA-*.csv）。

This dataset provides resources for training and evaluating SlideChat, a multimodal large language model for whole-slide pathology image understanding. It includes instruction-following training data and VQA/Caption evaluation benchmarks across multiple pathology cohorts and tasks. Specifically, the contents are: Training Instruction Data with Slide-level caption instruction data for Stage-1 training (SlideInstruct_train_stage1_caption.json) and Slide-level VQA instruction data for Stage-2 training (SlideInstruct_train_stage2_vqa.json); Evaluation Benchmarks (SlideBench) with Caption/Report evaluation splits for TCGA, CPTAC, and Histai cohorts (SlideBench-Caption-*.csv), and Closed VQA evaluation splits for TCGA, BCNB, CPTAC, and Histai cohorts (SlideBench-VQA-*.csv).