julyai/ProJudge-173k
收藏官方服务：
资源简介：
ProJudge-173k是一个大规模的指令微调数据集，专为基于MLLM的过程评估设计，包含细粒度的步骤级别注释。该数据集具有多模态特性，包含纯文本、单张图像和多张图像混合内容；跨学科特性，涵盖数学、物理、化学和生物学四个领域；并且具有从小学到竞赛级别的多难度级别。
ProJudge-173k is a large-scale instruction tuning dataset specifically designed for process evaluation based on MLLM, containing fine-grained step-level annotations. The dataset features multi-modality, including pure text, single image, and multi-image interleaved content; spans across four scientific disciplines: mathematics, physics, chemistry, and biology; and includes diverse difficulty levels from primary school to competition levels.
提供机构：
julyai
