Acerca de
The Equivariant Architecture Designer skill provides specialized guidance for building neural networks that respect physical or geometric symmetries, such as rotation, translation, and reflection. By leveraging domain-specific libraries like e3nn and escnn, it helps developers select the right architecture families, equivariant layers, and nonlinearities for tasks involving 3D point clouds, images, or molecular structures. This skill streamlines the complex process of building symmetry-aware models that generalize better with fewer parameters than standard deep learning approaches.
