A Particle Transformer jet tagger contains a sparse six-head circuit whose source-relay-readout structure recovers most performance and whose residual stream preferentially encodes 2-prong energy correlators.
hub Tool reference
Particle Transformer for Jet Tagging
Tool reference. 78% of classified Pith citations use this work as a method, library, or software dependency, not as a substantive claim.
31
Pith papers citing it
Method reference
78% of classified citations
abstract
Jet tagging is a critical yet challenging classification task in particle physics. While deep learning has transformed jet tagging and significantly improved performance, the lack of a large-scale public dataset impedes further enhancement. In this work, we present JetClass, a new comprehensive dataset for jet tagging. The JetClass dataset consists of 100 M jets, about two orders of magnitude larger than existing public datasets. A total of 10 types of jets are simulated, including several types unexplored for tagging so far. Based on the large dataset, we propose a new Transformer-based architecture for jet tagging, called Particle Transformer (ParT). By incorporating pairwise particle interactions in the attention mechanism, ParT achieves higher tagging performance than a plain Transformer and surpasses the previous state-of-the-art, ParticleNet, by a large margin. The pre-trained ParT models, once fine-tuned, also substantially enhance the performance on two widely adopted jet tagging benchmarks. The dataset, code and models are publicly available at https://github.com/jet-universe/particle_transformer.
hub tools
citation-role summary
method 6
background 2
dataset 1
citation-polarity summary
representative citing papers
PLuM multimodal transformer improves top and H->bb jet tagging by jointly processing particle constituents and Lund plane splittings, yielding 25% higher background rejection at 25% di-Higgs efficiency.
PHAT-JeT combines geometric message-passing with hierarchical patch attention to reach state-of-the-art accuracy and background rejection among resource-constrained jet tagging models on four benchmarks.
LP2B encoding converts Lund plane jet representations into Bloch sphere qubit states, enabling a QTTN that matches classical LundNet performance on polarization tagging and W/top tagging with three orders of magnitude fewer parameters and improved low-data regime results.
IAFormer uses boost-invariant pairwise quantities and differential attention to create a sparse Transformer that achieves state-of-the-art classification on top-quark and quark-gluon jet datasets while using over an order of magnitude fewer parameters than prior Particle Transformer models.
Quantity-grounded multi-agent decomposition makes LLM-generated collider analysis code inspectable and reliable with 14B-scale models, outperforming prior single-prompt approaches.
Point-cloud ML models classify charm- and bottom-origin electrons at ~80% purity for 40% efficiency, outperforming a BDT baseline, with performance limited by intrinsic decay similarity.
LLM embeddings condition a generative transformer to enable faster convergence, better performance, and generalization to unseen LHC processes using a single model.
Transformer models trained on FCC-ee CLD simulation reconstruct hadronic taus with per-mille mis-ID, F1 up to 0.95, sub-per-mille charge errors, and percent-level transverse momentum resolution.
Collider events are represented as multivectors in Cl(1,3) ⊗ V_flav whose grade projections recover standard observables, intended as input for equivariant foundation models.
Neural networks for HEP tasks can be fooled at significant rates by subtle perturbations inside uncertainty envelopes, revealing hidden systematics not captured by conventional methods.
Semi-leptonic h→VV* decays retain an effective two-qutrit quantum description under NLO QCD and electroweak corrections, unlike the fully leptonic h→4ℓ channel.
PaRT achieves >50% tagging efficiency for boosted H->WW jets at 1% background efficiency, decorrelated from jet mass, with data-to-simulation scale factors of 0.9-1.0 on 138 fb^{-1} of 13 TeV collisions.
Higgsformer achieves AUC 0.855 on t tbar H vs t tbar classification from raw hits, matching a Delphes-based Particle Transformer at ~40% b-tagging efficiency.
Equivariant jet taggers suppress frame-dependent pseudorapidity while encoding jet mass and N-subjettiness strongly, with bivector channels negligible and vector channels dominant for top tagging.
Presents a reusable open-source framework for mapping quantized transformer layers to AMD Versal AI Engine tiles for jet tagging at the LHC.
Release of an AI-ready dataset containing approximately 660,000 reconstructed polarized e+e- collision events at 91.2 GeV from the SLD experiment, translated from legacy formats with accompanying digitized documentation.
A hyper-graph neural network improves discrimination of four-top production at 13 TeV, raising expected significance from 5.13 to 9.11 and enabling projected 95% CL limits on five dimension-six SMEFT Wilson coefficients at current and HL-LHC luminosities.
Explainability techniques applied to LundNet show that assigned node importance correlates with classical jet substructure observables such as N-subjettiness ratios and energy correlation functions, with shifts across transverse-momentum regimes.
An edge-weighted multi-graph GNN (E-PCN/KIGNet) that encodes four jet kinematic variables improves JetClass jet-tagging accuracy and attributes most predictions to angular separation and transverse momentum.
SAL-T enhances the linformer with spatially aware kinematic partitioning and convolutions to match full-attention transformer performance on jet tagging while keeping linear complexity and lower latency.
Deep learning on all particles via holistic analysis and Advanced Color Singlet Identification improves Higgs signal extraction up to sixfold in high-energy collisions.
Transformers trained on cosmic ray simulations learn physically plausible features in positional encodings for symmetric air showers and in attention mechanisms for galaxy-origin particles.
Adversarial training enhances robustness of jet tagging classifiers while preserving performance, with loss surface geometry providing insights into correlations and vulnerability.
citing papers explorer
-
Dissecting Jet-Tagger Through Mechanistic Interpretability
A Particle Transformer jet tagger contains a sparse six-head circuit whose source-relay-readout structure recovers most performance and whose residual stream preferentially encodes 2-prong energy correlators.
-
Lund Plane to Bloch (LP2B) Encoding for Object and Polarization Tagging with Quantum Jet Substructure
LP2B encoding converts Lund plane jet representations into Bloch sphere qubit states, enabling a QTTN that matches classical LundNet performance on polarization tagging and W/top tagging with three orders of magnitude fewer parameters and improved low-data regime results.
-
One Generator, Any Process: LLM-Conditioning for the LHC
LLM embeddings condition a generative transformer to enable faster convergence, better performance, and generalization to unseen LHC processes using a single model.
-
Quantum Tomography and Entanglement in Semi-Leptonic $h\to VV^*$ Decays at Higher Orders
Semi-leptonic h→VV* decays retain an effective two-qutrit quantum description under NLO QCD and electroweak corrections, unlike the fully leptonic h→4ℓ channel.
-
Particle transformers for identifying Lorentz-boosted Higgs bosons decaying to a pair of W bosons
PaRT achieves >50% tagging efficiency for boosted H->WW jets at 1% background efficiency, decorrelated from jet mass, with data-to-simulation scale factors of 0.9-1.0 on 138 fb^{-1} of 13 TeV collisions.
-
Improved results on Higgs boson pair production in the 4b final state
CMS sets an observed upper limit of 4.4 on the HH signal strength μ_HH in the 4b final state at 13.6 TeV, improving prior LHC results by more than a factor of two in the resolved topology.