Reproduce “Learning the Cost of Reliable 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
Dimitrios Rontogiannis, Ander Artola Velasco, Manuel Gomez Rodriguez
The authors propose a model-routing platform that learns which providers meet a user’s average-quality threshold and uses competitive bids to allocate queries. It explores to estimate quality and cost, then pays providers a critical price. In theory, average payments approach the second-lowest mean cost among qualified providers; simulations show payments below listed prices in GPQA but above them in GSM8K.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.