However, developing models that can process weight-space objects is challenging due to their high dimensional nature.

We integrate this model in a planning loop, where conditioning and classifier guidance let us softly break the symmetry for specific tasks as needed.

It is common that these data themselves have inherently complicated 3D geometric structures.

It is common that these data themselves have inherently complicated 3D geometric structures.

We show here that this approach can effectively be used to make a large pretrained network equivariant.

However, for many applications this formulation is insufficient.

Such data can take numerous forms, for instance points, direction vectors, translations, or rotations, but to date there is no single architecture that can be applied to such a wide variety of geometric types while respecting their symmetries. In this paper we introduce the Geometric Algebra Transformer (GA Tr), a general-purpose architecture for geometric data.