Extrapolating Jet Radiation with Autoregressive Transformers

Anja Butter, François Charton, Javier Mariño Villadamigo, Ayodele Ore, Tilman Plehn +1 more
12/16/2024
hep-phcs.LG

No official code found. Best implementation: heidelberg-hepml/jetgpt-splittings (1 star, last commit 2024-12-29).

Abstract

Generative networks are an exciting tool for fast LHC event fixed number of particles. Autoregressive transformers allow us to generate events containing variable numbers of particles, very much in line with the physics of QCD jet radiation, and offer the possibility to generalize to higher multiplicities. We show how transformers can learn a factorized likelihood for jet radiation and extrapolate in terms of the number of generated jets. For this extrapolation, bootstrapping training data and training with modifications of the likelihood loss can be used.

View on arXivView PDF

Code Implementations(1)

heidelberg-hepml/jetgpt-splittingsCommunity100%

Repository for <Extrapolating Jet Radiation with Autoregressive Transformers> (A. Butter et al 2024)

10PythonDec 16, 20241 years ago

Similar Papers

Forecasting Generative Amplification

80% match2025

Stochastic Latent Transformer: Efficient Modelling of Stochastically Forced Zonal Jets

80% match2023

Interpreting Transformers for Jet Tagging

78% match1 repo32024

Transformers Simulate MLE for Sequence Generation in Bayesian Networks

78% match2025

Sub-microsecond Transformers for Jet Tagging on FPGAs

77% match2025

Multimodal Generative Flows for LHC Jets

77% match2025

Cite this paper

@article{butter2024extrapolating,
  title  = {Extrapolating Jet Radiation with Autoregressive Transformers},
  author = {Anja Butter and François Charton and Javier Mariño Villadamigo and Ayodele Ore and Tilman Plehn and Jonas Spinner},
  year   = {2024},
  eprint = {2412.12074},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2412.12074v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2412.12074.svg)](https://codeofpaper.com/paper/2412.12074)

Discussion