Reproduce “Scaling Up Bayesian DAG Sampling” 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

Daniele Nikzad, Alexander Zhilkin, Juha Harviainen, Jack Kuipers, Giusi Moffa, Mikko Koivisto

Gibby approximates posterior distributions over Bayesian network structures with faster single-arc sampling and pruning parent-set scores for broader resampling moves. It makes repeated self-transitions cheaper to simulate while preserving the basic-move chain, and its pruning rule bounds errors in targeted score sums. On PIGS, its basic-move implementation achieved about 200 times GC’s throughput, not better mixing or posterior accuracy.

Summary by alphaXiv.

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