Gerasimov et al . claim that the ability of DM21 to respect fractional charge (FC) and fractional spin (FS) conditions outside of the training set has not been demonstrated in our paper. This is based on (i) asserting that the training set has a ~50% overlap with our bond-breaking benchmark (BBB) and (ii) questioning the validity and accuracy of our other generalization examples. We disagree with their analysis and believe that the points raised are either incorrect or not relevant to the main conclusions of the paper and to the assessment of general quality of DM21.

中文翻译：

---

回应评论“通过解决分数电子问题推动密度泛函的前沿”

格拉西莫夫等人. 声称 DM21 尊重训练集之外的分数电荷 (FC) 和分数自旋 (FS) 条件的能力尚未在我们的论文中得到证明。这是基于 (i) 断言训练集与我们的断键基准 (BBB) 有约 50% 的重叠，以及 (ii) 质疑我们其他泛化示例的有效性和准确性。我们不同意他们的分析，并认为提出的观点要么不正确，要么与论文的主要结论以及对 DM21 总体质量的评估无关。