Reproduce “tfp.mcmc: Modern Markov Chain Monte Carlo Tools Built for Modern Hardware” 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
Junpeng Lao, Christopher Suter, Ian Langmore, Cyril Chimisov, Ashish Saxena, Pavel Sountsov, Dave Moore, Rif A. Saurous and 2 more
TFP’s MCMC toolkit supports parallel Bayesian sampling by treating chain states as batches that models and transition kernels process together. Its callable target log probability need not be normalized. For NUTS, TFP uses a TensorFlow loop and preallocated storage, reducing state-and-momentum-history memory from O(2^max_tree_depth) to O(max_tree_depth). This is an algorithmic design claim, not a measured speedup.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.