EquiformerV2: Improved Equivariant Transformer for Scaling to Higher-Degree Representations

TL;DR

EquiformerV2 advances equivariant Transformer architectures for 3D atomistic systems by scaling to higher degrees, improving efficiency, accuracy, and data utilization, and demonstrating superior performance on multiple large-scale datasets.

Contribution

The paper introduces EquiformerV2, a scalable and efficient equivariant Transformer architecture with novel architectural improvements for higher-degree representations.

Findings

Outperforms previous state-of-the-art on OC20 dataset

Achieves up to 9% better force prediction accuracy

Reduces DFT calculations by 2x for adsorption energies

Abstract

Equivariant Transformers such as Equiformer have demonstrated the efficacy of applying Transformers to the domain of 3D atomistic systems. However, they are limited to small degrees of equivariant representations due to their computational complexity. In this paper, we investigate whether these architectures can scale well to higher degrees. Starting from Equiformer, we first replace $SO(3)$ convolutions with eSCN convolutions to efficiently incorporate higher-degree tensors. Then, to better leverage the power of higher degrees, we propose three architectural improvements -- attention re-normalization, separable $S^2$ activation and separable layer normalization. Putting this all together, we propose EquiformerV2, which outperforms previous state-of-the-art methods on large-scale OC20 dataset by up to $9\%$ on forces, $4\%$ on energies, offers better speed-accuracy trade-offs, and $2\times$ reduction in DFT calculations needed for computing adsorption energies. Additionally, EquiformerV2 trained on only OC22 dataset outperforms GemNet-OC trained on both OC20 and OC22 datasets, achieving much better data efficiency. Finally, we compare EquiformerV2 with Equiformer on QM9 and OC20 S2EF-2M datasets to better understand the performance gain brought by higher degrees.