AG-2025.12-1732·hep-ph·cross-listed: hep-ex
Economical Jet Taggers -- Equivariant, Slim, and Quantized
Authors
- Antoine Petitjean
- Tilman Plehn
- Jonas Spinner
- Ullrich Köthe
Abstract
Modern machine learning is transforming jet tagging at the LHC, but the leading transformer architectures are large, not particularly fast, and training-intensive. We present a slim version of the L-GATr tagger, reduce the number of parameters of jet-tagging transformers, and quantize them. We compare different quantization methods for standard and Lorentz-equivariant transformers and estimate their gains in resource efficiency. We find an order-of-magnitude reduction in energy cost for an moderate performance decrease, down to 1000-parameter taggers. This might be a step towards trigger-level jet tagging with small and quantized versions of the leading equivariant transformer architectures.
Submitted
18 December 20254 months ago
Version
v1
License
CC-BY-4.0
DOI
10.48550/arXiv.2512.17011
Chat with this PDF
Ask questions, probe assumptions, request a plain-English summary. Answers cite sections from the preprint itself.
Community
Questions and answers about this paper from other readers. No formal peer review — just a place to think out loud.