CMS Level-1 Trigger Anomaly Detection Dataset (2025 Zero-Bias Data and Simulations)
- Provider
- CMS Level-1 Trigger anomaly detection group, CERN
- License
- CC0-1.0 (source)
- Size
- 152 files
- Format
- Parquet (in sampled files)
- Stored at
- Hugging Face (https://huggingface.co/datasets/CERN/anomaly_detection_cmsl1t)
- Access
- Free and public. No account with the data provider is needed.
- DOI
- 10.5281/zenodo.21787779
- Category
- Physics
- Source
- https://doi.org/10.5281/zenodo.21787779
About this dataset
Level-1 Trigger objects from the CMS experiment at the CERN Large Hadron Collider, assembled for research on unsupervised anomaly detection in the trigger as a route to new physics searches. The normal data are 20,887,636 zero-bias events recorded in 2025 proton-proton running (runs 396102 and 398183), chosen at random from all collisions. Twenty simulated signal sets from the CMS Run 3 Winter25 campaign (Higgs, multi-Higgs, SUSY and exotic scenarios, 13,012,931 events) and a 2,000,000 event SingleNeutrino simulation of the normal data are included. The publishers state the data contain no new physics. Features are in trigger-native hardware integer format, pre-split into train, valid and test. Listing data/ (measured 2026-11-13) shows 23 folders, one per data set (for example ZB_run396102, HHHto6B, WtoTauto3Mu, SUSYGluGluToBBHToBB), each holding Parquet shards plus a seeds/ folder with the decisions of the standard L1 menu algorithms for comparison (the AXO and CICADA anomaly triggers are left out). loader/ and configs/ hold a PyTorch dataloader. Open with Hugging Face datasets (one configuration per data set), pandas or pyarrow. A mirror of Zenodo record 10.5281/zenodo.21787779, released under CC0-1.0; citation is requested as a courtesy.
How to load this dataset in Python
from datasets import load_dataset
ds = load_dataset("CERN/anomaly_detection_cmsl1t")
print(ds)Data structure
File types in sampled files: Parquet
Files
Example file: data/GluGluHTo2B_Par-MH-125/seeds/test-00000-of-00001.parquet
Analyze this dataset without downloading it
On Scigantic, CMS Level-1 Trigger Anomaly Detection Dataset (2025 Zero-Bias Data and Simulations) opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.
This page is also available as Markdown and as JSON.
Related Physics datasets
- ColliderML Drifting Beamspot: Single-Muon Tracking Samples with Smeared Digitisation
- Top Quark Tagging Reference Dataset (Parquet)
- Rotor37: 3D Compressor Blade RANS Dataset under Geometric Variations (PLAID)
- The Well: Euler Multi-Quadrant Riemann Problems (Open Boundaries)
- Pythia 8 Quark and Gluon Jets (Komiske, Metodiev, Thaler), Parquet with Particle Transformer Features
- The Well: Rayleigh-Taylor Instability Turbulence (3D)
- The Well: Acoustic Scattering Across a Discontinuous Interface
- DeepMind Hamiltonian Dynamics Suite: 17 Simulated Physical Systems for Learning Latent Dynamics