MassSpecGym

该数据集是迄今为止最大的公开标记质谱数据集，它汇集了来自多个公共参考光谱数据库的资源。此外，该数据集还提供了一个基于最大公共边缘子结构（MCES）的建议分割方式，以便对模型进行评估。在三个数据集中，它的规模是最大的。该数据集的任务是对分子生成模型进行基准测试和评估。

This dataset is the largest publicly labeled mass spectrometry dataset to date, aggregating resources from multiple public reference spectral databases. Furthermore, it provides a proposed data splitting strategy based on the Maximum Common Edge Substructure (MCES) to support model evaluation. It is the largest in scale among the three datasets. The core task of this dataset is to benchmark and evaluate molecular generative models.