New AI model TabPFN enables faster and more precise predictions on small table data sets

Filling gaps in data sets or identifying outliers - this is what the TabPFN machine learning algorithm developed by a team led by Prof. Dr. Frank Hutter can do. This artificial intelligence (AI) uses learning methods inspired by large language models. TabPFN learns causal relationships from synthetic data and is therefore more often correct with its predictions than previously standard algorithms. The results were published in the journal Nature.

“Being able to use TabPFN to reliably and quickly calculate predictions from tabular data is a benefit for many disciplines - from biomedicine to economics and physics,” says Hutter. “TabPFN delivers better results faster and is ideal for small companies and teams due to its low resource and data requirements.” Interested parties can find the code and instructions on how to use it at priorlabs.ai/tabpfn-nature/.

The original press release can be found on the university's website uni-freiburg.de/neues-ki-modell-tabpfn-ermoeglicht-schnellere-und-praezisere-vorhersagen-auf-kleinen-tabellendatensaetzen/

Publication: Hollmann, N., Müller, S., Purucker, L. et al. Accurate predictions on small data with a tabular foundation model. Nature 637, 319–326 (2025). doi.org/10.1038/s41586-024-08328-6