UKAEA-IBM-STFC/tokamark-dataset

TokaMark: A Comprehensive Benchmark for MAST Tokamak Plasma Models

Abstract: Development and operation of commercially viable fusion energy reactors such as tokamaks require accurate predictions of plasma dynamics from sparse, noisy, and incomplete sensors readings. The complexity of the underlying physics and the heterogeneity of experimental data pose formidable challenges for conventional numerical methods, while simultaneously highlights the promise of modern data-native AI approaches. A major obstacle in realizing this potential is, however, the lack of curated, openly available datasets and standardized benchmarks. Existing fusion datasets are scarce, fragmented across institutions, facility-specific, and inconsistently annotated, which limits reproducibility and prevents a fair and scalable comparison of AI approaches. In this paper, we introduce TokaMark, a structured benchmark to evaluate AI models on real experimental data collected from the Mega Ampere Spherical Tokamak (MAST). TokaMark provides a comprehensive suite of tools designed to (i) unify access to multi-modal heterogeneous fusion data (ii) harmonize formats, metadata, temporal alignment and evaluation protocols to enable consistent cross-model and cross-task comparisons. The benchmark includes a curated list of 14 tasks spanning a range of physical mechanisms, exploiting a variety of diagnostics and covering multiple target use cases. A baseline model is provided to facilitate transparent comparison and validation within a unified framework. By establishing a unified benchmark for both the fusion and AI-for-science communities, TokaMark aims to accelerate progress in data-driven plasma AI modeling, contributing to the broader goal of achieving sustainable and stable fusion energy. The benchmark, documentation, and tooling are fully open sourced to encourage community adoption and contribution.

Downloading Data

The data can be downloaded from the huggingface repo with the following command:

hf download UKAEA-IBM-STFC/tokamark-dataset --repo-type dataset --local-dir <my/local/path>

After the download is complete, you can extract the Zarr data files used for training and evaluation from the zip folders with the following script. Please be aware that the unzipped data is approximately ~2TB in size:

cd <my/local/path>
./unzip_dataset.sh

Citation Information

If you use TokaMark in your research, please cite:

@article{tokamark2026,
  title={TokaMark: A Comprehensive Benchmark for Learning Plasma Dynamics in MAST Tokamak},
  author={Cécile Rousseau,
, Samuel Jackson, Rodrigo H. Ordonez-Hurtado, Nicola C. Amorisco, Tobia Boschi, George K. Holt, Andrea Loreti, Eszter Szekely, Alexander Whittle, Adriano Agnello, Stanislas Pamela, Alessandra Pascale, Robert Akers, Juan Bernabe Moreno, H. Sue Thorne, Mykhaylo Zayats},
  journal={KDD 2026 Datasets and Benchmark Track},
  year={2026}
}

The data used in TokaMark is derved from the FAIR-MAST Data Catalog:

@article{jackson_open_2025,
  title={An Open Data Service for Supporting Research in Machine Learning on Tokamak Data},
  author={Jackson, Samuel and Khan, Saiful and Cummings, Nathan and Hodson, James and de Witt, Shaun and Pamela, Stanislas and Akers, Rob and Thiyagalingam, Jeyan},
  journal={IEEE Transactions on Plasma Science},
  year={2025},
  issn={1939-9375},
  doi={10.1109/TPS.2025.3583419}
}

@article{jackson2024fair,
  title={FAIR-MAST: A fusion device data management system},
  author={Jackson, Samuel and Khan, Saiful and Cummings, Nathan and Hodson, James and de Witt, Shaun and Pamela, Stanislas and Akers, Rob and Thiyagalingam, Jeyan and The MAST Team},
  journal={SoftwareX},
  volume={27},
  pages={101869},
  year={2024},
  publisher={Elsevier}
}