Reproduce “Automatic Construction of Strong Simple Baselines” 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
Caleb N. Ellington, Jiaqi Wang, Sohan Addagudi, Elijah Cole
The authors use a language model to build and run baseline predictors for 36 publication-derived scientific classification and regression tasks, comparing held-out scores with trivial predictors and published deep-learning results. Across task medians, baselines recovered a median 95.5% of the score gap; 12 of 36 met or exceeded the reference. Results concern one model and selected tasks.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.