{
  "access": "open",
  "card": {
    "intended_use": "Training and comparing reconstruction / classification models across detector geometries.",
    "known_limitations": "No DOI or licence; files hosted via ERDA share links listed in the GitHub README.",
    "out_of_scope_use": "Physics results: geometries are inspired by, not identical to, real detectors; no detector data.",
    "summary": "Cross-experiment ML benchmark for neutrino-telescope reconstruction."
  },
  "channels": ["cc", "nc"],
  "cite_as": ["arXiv:2511.13111"],
  "creators": [
    {
      "name": "Rasmus F. Ørsøe"
    },
    {
      "name": "Stephan Meighen-Berger"
    },
    {
      "name": "Jeffrey Lazar"
    },
    {
      "name": "Jorge Prado"
    },
    {
      "name": "Ivan Mozun-Mateo"
    },
    {
      "name": "Aske Rosted"
    },
    {
      "name": "Philip Weigel"
    },
    {
      "name": "Arturo Llorente Anaya"
    }
  ],
  "curation": {
    "added": "2026-10-23",
    "quality_notes": ["the per-dataset ERDA share links are in the README; n_events is the approximate total from the abstract"],
    "source": "manual (user suggestion)",
    "status": "curated"
  },
  "data": {
    "formats": ["parquet"],
    "has_mc": true,
    "levels": ["mc"],
    "n_events": 130000000,
    "observables": ["pulses", "event-level truth"]
  },
  "date_published": "2025-11",
  "description": "Seven large-scale simulated datasets with nearly 130 million charged- and neutral-current muon-neutrino\ninteractions from 10 GeV to 100 TeV, generated for six detector geometries inspired by existing and proposed\nneutrino telescopes in water and ice (\"Cluster\", \"Flower S/L/XL\", \"Hexagon\", \"Hexagon Ice LE\", \"Triangle\").\nPulse- and event-level information for energy and direction reconstruction, topology classification, vertex and\ninelasticity estimation, with train/test selections, the predictions of four reference models (ParticleNeT,\nDynEdge, GRIT, DeepIce) and their trained model artifacts. Compatible with the GraphNeT dataset classes.",
  "detector_technologies": ["ice-cherenkov", "water-cherenkov-open"],
  "distributions": [
    {
      "host": "github",
      "label": "catalogue and download links",
      "role": "documentation",
      "url": "https://github.com/graphnet-team/NuBench"
    },
    {
      "formats": ["parquet", "tarball"],
      "host": "institutional",
      "label": "ERDA (University of Copenhagen): per-dataset SQLite / Parquet, predictions and model artifacts (tar.gz)",
      "role": "primary",
      "url": "https://sid.erda.dk/"
    }
  ],
  "energy_range": {
    "max": 100000,
    "min": 10,
    "quantity": "true neutrino energy",
    "unit": "GeV"
  },
  "flavors": ["numu", "numubar"],
  "id": "nubench",
  "license": "none-stated",
  "owner": {
    "kind": "organization",
    "name": "graphnet-team"
  },
  "references": [
    {
      "arxiv": "2511.13111",
      "citation": "Ørsøe et al., NuBench: An Open Benchmark for Deep Learning-Based Event Reconstruction in Neutrino Telescopes (prepared for JINST)",
      "relation": "describes"
    }
  ],
  "relations": [
    {
      "notes": "simulation cited in the catalogue README",
      "rel": "derived_from",
      "target": "prometheus"
    },
    {
      "rel": "uses",
      "target": "graphnet"
    }
  ],
  "short": "NuBench: 7 simulated neutrino-telescope datasets (~130 M νμ events, 6 geometries) for ML reconstruction",
  "sources": ["atmospheric", "astrophysical"],
  "title": "NuBench: open benchmark datasets for deep-learning event reconstruction in neutrino telescopes",
  "topics": ["detector-response", "statistics"],
  "type": "dataset",
  "url_landing": "https://github.com/graphnet-team/NuBench"
}