arXiv · 2608.02735
Virtues and Vices of Equivariant Transformers
Abstract
We study for the first time the benefit of Lorentz-equivariant transformers for large-size jet tagging and flavor tagging. To control their computing demands, we optimize all implementations for inference cost metrics. In our scaling studies, we find that Lorentz-equivariant networks outperform standard transformers, provided geometric features are relevant. This holds true in an idealized world as well as for limited resources. The conditional gain from Lorentz equivariance provides interesting input to the development of foundation models for LHC data.
Explore related subjects
Keep this discovery
Luigi Favaro, Tilman Plehn, Huilin Qu, Jonas Spinner. 2026-08-03. Virtues and Vices of Equivariant Transformers. https://arxiv.org/abs/2608.02735
Cite the original work for its findings. Save a collection to share your selection of sources.