ColabFit

The ColabFit Exchange: Data for Advanced Materials Science

Spotlight at NeurIPS AI4Mat Workshop 2024

December 2024:
E. G. Fuemmeler, G. P. Wolfe, A. Gupta, J. A. Vita, E. B. Tadmor, S. Martiniani, "Advancing the ColabFit Exchange towards a Web-scale Data Source for Machine Learning Interatomic Potentials", NeurIPS, (2024). https://openreview.net/forum?id=b8qZpGJIkw

ColabFit Paper

Published October 2023:
J. A. Vita, E. G. Fuemmeler, A. Gupta, G. P. Wolfe, A. Q. Tao, R. S. Elliott, S. Martiniani, E. B. Tadmor, "ColabFit exchange: Open-access datasets for data-driven interatomic potentials", Journal of Chemical Physics, 159, 154802 (2023). doi:10.1063/5.0163882

@article{10.1063/5.0163882,
    author = {Vita, Joshua A. and Fuemmeler, Eric G. and Gupta, Amit and Wolfe, Gregory P. and Tao, Alexander Quanming and Elliott, Ryan S. and Martiniani, Stefano and Tadmor, Ellad B.},
    title = "{ColabFit exchange: Open-access datasets for data-driven interatomic potentials}",
    journal = {The Journal of Chemical Physics},
    volume = {159},
    number = {15},
    pages = {154802},
    year = {2023},
    month = {10},
    abstract = "{Data-driven interatomic potentials (IPs) trained on large collections of first principles calculations are rapidly becoming essential tools in the fields of computational materials science and chemistry for performing atomic-scale simulations. Despite this, apart from a few notable exceptions, there is a distinct lack of well-organized, public datasets in common formats available for use with IP development. This deficiency precludes the research community from implementing widespread benchmarking, which is essential for gaining insight into model performance and transferability, and also limits the development of more general, or even universal, IPs. To address this issue, we introduce the ColabFit Exchange, the first database providing open access to a large collection of systematically organized datasets from multiple domains that is especially designed for IP development. The ColabFit Exchange is publicly available at https://colabfit.org, providing a web-based interface for exploring, downloading, and contributing datasets. Composed of data collected from the literature or provided by community researchers, the ColabFit Exchange currently (September 2023) consists of 139 datasets spanning nearly 70,000 unique chemistries, and is intended to continuously grow. In addition to outlining the software framework used for constructing and accessing the ColabFit Exchange, we also provide analyses of the data, quantifying the diversity of the database and proposing metrics for assessing the relative diversity of multiple datasets. Finally, we demonstrate an end-to-end IP development pipeline, utilizing datasets from the ColabFit Exchange, fitting tools from the KLIFF software package, and validation tests provided by the OpenKIM framework.}",
    issn = {0021-9606},
    doi = {10.1063/5.0163882},
    url = {https://doi.org/10.1063/5.0163882},
    eprint = {https://pubs.aip.org/aip/jcp/article-pdf/doi/10.1063/5.0163882/18177819/154802\_1\_5.0163882.pdf},
}

The ColabFit Exchange is a KIM Initiative project. It is the largest database of its kind, providing open access to a large collection of heterogeneous but systematically organized datasets that are especially designed for data-driven interatomic potential (DDIP) development.

Each curated dataset is represented as data points with selected training features and targets, and can be individually downloaded as a set of tarred parquet files (see the documentation here for guidance on using the parquet files). Datasets are additionally exported as extended XYZ files, which can be read by, for example, the ASE Python library.

Datasets can also be accessed on the ColabFit organization page on HuggingFace, via the ColabFit CLI, or through the KLIFF DDIP fitting package.
The ColabFit MCP (Model Context Protocol) is also available to allow AI/LLM interactions with the ColabFit Exchange.

The ColabFit Exchange, along with the other KIM Initiative projects, addresses a pressing need in the molecular simulation community—ensuring science is findable, accessible, interoperable, and reproducible.