vector-institute/atom3d-smp

QM9数据集包含134k个稳定的小有机分子的结构和能量、电子及热力学性质，这些数据来源于量子化学计算。数据集的任务是从基态结构预测分子性质。数据集提供了一个随机分割的训练集、验证集和测试集。部分分子因未通过一致性测试或未正确处理而被排除。labels字段包含了每个分子的旋转常数、偶极矩、各向同性极化率、HOMO能量、LUMO能量、能隙、电子空间范围、零点振动能量、内能、焓、自由能、热容等量及其单位。

The QM9 dataset contains structures and energetic, electronic, and thermodynamic properties for 134k stable small organic molecules, obtained from quantum-chemical calculations. The task is to predict the molecular properties from the ground-state structure. The dataset is randomly split into training, validation, and test sets. The labels field contains multiple physical quantities for each item, such as rotational constants, dipole moments, polarizabilities, etc.