GLOW: A Unified Particle Flow Transformer

Dmitrii Kobylianskii; Samuel Van Stroud; Kwok Yiu Wong; Max Hart; Etienne Dreyer; Eilam Gross; Gabriel Facini; Tim Scanlon

doi:10.48550/arXiv.2508.20092

GLOW: A Unified Particle Flow Transformer

Dmitrii Kobylianskii
, Samuel Van Stroud
, Kwok Yiu Wong
, Max Hart
, Etienne Dreyer
, Eilam Gross
, Gabriel Facini
, Tim Scanlon

Department of Particle Physics and Astrophysics

Weizmann Institute of Science

Research output: Contribution to journal › Article

Abstract

We present GLOW, a transformer-based particle flow model that combines incidence matrix supervision from HGPflow with a MaskFormer architecture. Evaluated on CLIC detector simulations, GLOW achieves state-of-the-art performance and, together with prior work, demonstrates that a single unified transformer architecture can effectively address diverse reconstruction tasks in particle physics.

Original language	English GB
Journal	arxiv.org
DOIs	https://doi.org/10.48550/arXiv.2508.20092
State	In preparation - 27 Aug 2025

Access to Document

10.48550/arXiv.2508.20092License: CC BY

Cite this

@article{e8a0fa5de1db4eb7abe79c694f1b93f6,

title = "GLOW: A Unified Particle Flow Transformer",

abstract = "We present GLOW, a transformer-based particle flow model that combines incidence matrix supervision from HGPflow with a MaskFormer architecture. Evaluated on CLIC detector simulations, GLOW achieves state-of-the-art performance and, together with prior work, demonstrates that a single unified transformer architecture can effectively address diverse reconstruction tasks in particle physics.",

author = "Dmitrii Kobylianskii and Stroud, \{Samuel Van\} and Wong, \{Kwok Yiu\} and Max Hart and Etienne Dreyer and Eilam Gross and Gabriel Facini and Tim Scanlon",

year = "2025",

month = aug,

day = "27",

doi = "10.48550/arXiv.2508.20092",

language = "الإنجليزيّة",

}

TY - JOUR

T1 - GLOW: A Unified Particle Flow Transformer

AU - Kobylianskii, Dmitrii

AU - Stroud, Samuel Van

AU - Wong, Kwok Yiu

AU - Hart, Max

AU - Dreyer, Etienne

AU - Gross, Eilam

AU - Facini, Gabriel

AU - Scanlon, Tim

PY - 2025/8/27

Y1 - 2025/8/27

N2 - We present GLOW, a transformer-based particle flow model that combines incidence matrix supervision from HGPflow with a MaskFormer architecture. Evaluated on CLIC detector simulations, GLOW achieves state-of-the-art performance and, together with prior work, demonstrates that a single unified transformer architecture can effectively address diverse reconstruction tasks in particle physics.

AB - We present GLOW, a transformer-based particle flow model that combines incidence matrix supervision from HGPflow with a MaskFormer architecture. Evaluated on CLIC detector simulations, GLOW achieves state-of-the-art performance and, together with prior work, demonstrates that a single unified transformer architecture can effectively address diverse reconstruction tasks in particle physics.

U2 - 10.48550/arXiv.2508.20092

DO - 10.48550/arXiv.2508.20092

M3 - مقالة

SN - 2331-8422

JO - arxiv.org

JF - arxiv.org

ER -

GLOW: A Unified Particle Flow Transformer

Abstract

Access to Document

Fingerprint

Cite this