Are Scaling Laws of Universal MLIPs Universal?

ML NNGTB GNN and MP GNNs AI4S IMMPP
神经尺度定律可用于预测增大训练数据集所带来的性能提升,但机器学习原子间势(MLIP)的数据集具有高度异质性,且生成成本高昂;相同数量的结构未必蕴含等量的有效信息。本文在固定模型配置的前提下,研究了M3GNet、MACE和PET三种模型在MatPES-MP-ALOE联合训练池子集上的数据规模扩展规律。实验覆盖了从10,000至640,000个训练结构的多个子集,并包含全部含116万个结构的完整训练集;对比了五种采样策略:均匀随机采样、三种基于化学组成与结构特征的分层采样策略、迭代式DIRECT采样,以及利用三个预训练MLIP模型所提取嵌入表示进行的最远点采样(furthest-point sampling)。模型评估采用内聚能、原子力、应力及综合损失四项指标,并分别在两个测试集上进行:一是从原始母分布中独立留出的测试集,二是分布发生偏移的MP-r²SCAN测试集。我们观察到,随着训练数据量增加,各项指标普遍呈现幂律形式的性能提升;但拟合所得的幂律指数、表观误差下限(apparent error floor)以及相对采样效率,均显著依赖于模型架构、预测目标量(如能量、力或应力)、数据采样策略,以及测试集所属的分布类型。值得注意的是,不同采样策略的优劣排序会随测试分布的改变而发生反转:例如,基于化学组成的分层采样可能提升跨数据集泛化能力,却同时损害对原始分布的拟合精度;而基于嵌入空间多样性的采样策略则往往表现不如随机采样。此外,模型在测试集上的相对损失与测试集和训练集在化学空间中的偏离程度呈强相关性。上述结果表明,结构数量虽是MLIP规模扩展中不可或缺的变量,却远非充分变量。在训练MLIP时,额外开展的第一性原理计算所能带来的收益,不仅取决于新增结构的数量,更关键地取决于这些结构在化学组成与原子构型上的内容,是否契合该势函数模型预期部署的具体应用场景。
Neural scaling laws enable the prediction of returns from larger training datasets, but machine-learned interatomic potential (MLIP) datasets are heterogeneous and expensive to generate, and equal numbers of structures need not contain equal amounts of useful information. We study data scaling under fixed model configurations for M3GNet, MACE, and PET trained on subsets of a combined MatPES-MP-ALOE training pool. Across training structure subsets from 10,000 to 640,000, together with the full 1.16-million-structure training set, we compare uniform random sampling, three chemistry- and structure-informed stratification policies, iterative DIRECT, and furthest-point sampling using embeddings from three pretrained MLIPs. Models are evaluated using cohesive-energy, force, stress, and aggregate losses on both a held-out test set from the parent distribution and the distribution shifted MP-r$^2$SCAN test set. We observe power-law scaling improvements with increasing data, but the fitted scaling exponent, apparent error floor, and relative sample efficiency depend on the model architecture, target quantity, data sampling policy, and test set distribution. Sampling-policy rankings can reverse across test distributions: chemical stratification can improve cross-dataset performance while degrading in-distribution performance, whereas the embedding-space diversity policies frequently under-perform random sampling. Relative loss is strongly associated with test-to-train chemical-space divergence. These results show that structure count is a necessary but insufficient scaling variable for MLIPs. For training MLIPs, the return on additional first-principles calculations depends not only on how many additional structures are included, but also on how their chemical and configurational content aligns with the model's intended use.
许愿