24–28 Aug 2026
Kirchhoff Institute for Physics (KIP)
Europe/Berlin timezone

Economical Jet Taggers -- Equivariant, Slim and Quantized

27 Aug 2026, 14:10
8m
3.404

3.404

Real-time Data Processing 🔀 Real-Time Data Processing

Speaker

Antoine Petitjean (ITP, Heidelberg University)

Description

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.

Authors

Antoine Petitjean (ITP, Heidelberg University) Jonas Spinner Tilman Plehn (Institut für Theoretische Physik, Universität Heidelberg; Interdisciplinary Center for Scientific Computing (IWR), Universität Heidelberg) Prof. Ullrich Köthe (IWR, Universität Heidelberg)

Presentation materials