Reproduce “Large-scale empirical tuning and comparison of default optimizers for variational inference” 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

Trevor Campbell, Jonathan H. Huggins, Kyurae Kim, Charles C. Margossian

The authors benchmark default optimizers for black-box variational inference across 1,092 problems, using 56 algorithms and selecting the best estimated objective from five independently run methods. The panel performs reliably near the best observed methods. Adam at 10^-4 ranks first on about 5–8% of problems and in the top 10 about 35%; results cover this suite, not all inference problems.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.