GDL-DS
收藏数据链接:
官方服务:
资源简介:
GDL-DS是一个综合基准,用于评估在分布偏移场景下的几何深度学习模型性能。评估数据集涵盖了从粒子物理学、材料科学到生物化学的多个科学领域,并包含了包括条件偏移、协变量偏移和概念偏移在内的广泛分布偏移。
GDL-DS is a comprehensive benchmark for evaluating the performance of geometric deep learning models under distribution shift scenarios. The benchmark dataset covers multiple scientific domains ranging from particle physics, materials science to biochemistry, and includes a wide range of distribution shifts such as conditional shift, covariate shift and conceptual shift.
创建时间:
2023-10-13



