Reproduce “Beyond Foundation Models: PFN Priors Improve Tabular Hyperparameter Optimization” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Artem Chistyakov, Renat Sergazinov, Artem Babenko
PriorTune learns a reusable shortlist of configurations for conventional tabular models from synthetic tasks generated by a tabular foundation-model prior. It chooses configurations against same-size random subsets, then fits only those configurations on new datasets. On BeyondArena, LightGBM with TabPrep features ranks seventh of 58 methods at 1313 Elo and 4.4 ms per row; six higher-ranked methods are pretrained foundation models.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.