Reproduce “CheMixHub: Datasets and Benchmarks for Chemical Mixture Property Prediction” 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

Ella Miray Rajaonson, Mahyar Rajabi Kochi, Luis Martin Mejia Mendoza, Seyed Mohamad Moosavi, Benjamin Sanchez-Lengeling

CheMixHub is a benchmark with 11 regression tasks from 7 datasets and about 500,000 data points. It evaluates under four data splits. On four temperature-dependent tasks, an Arrhenius-based head lowers MAE on held-out temperature-range splits; for NIST log-viscosity, MAE falls from 0.079 ± 0.006 to 0.048 ± 0.013. This supports the head under these tested conditions, not general mixture-design success.

Summary by alphaXiv.

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