AlgoTune: Can Language Models Speed Up General-Purpose Numerical Programs?

Ori Press, Brandon Amos, Haoyu Zhao, Yikai Wu, Samuel Ainsworth, Dominik Krupke, Patrick Kidger, Touqir Sajed, Bartolomeo Stellato, Jisun Park, Nathanael Bosch, Eli Meril, Albert Steppi, Arman Zharmagambetov, Fangzhao Zhang, David Pérez-Piñeiro, Alberto Mercurio, Ni Zhan, Talor Abramovich, Kilian Lieret, Hanlin Zhang, Shirley Huang, Matthias Bethge, Ofir Press

University of Tübingen· Princeton· Meta· TU Braunschweig· Cradle Bio· LG Electronics Canada· Seoul National University· Tel Aviv University· Quansight PBC· Stanford· Norwegian University of Science and Technology· EPFL· Harvard

language models benchmark program synthesis

⋅ NeurIPS ⋅ Project Page ⋅Poster ⋅OpenReview

Abstract

Despite progress in language model (LM) capabilities, evaluations have thus far focused on models' performance on tasks that humans have previously solved, including in programming (SWE-Bench) and mathematics (FrontierMath). We therefore propose testing models' ability to design and implement algorithms in an open-ended benchmark: We task LMs with writing code that efficiently solves computationally challenging problems in computer science, physics, and mathematics.

Our AlgoTune benchmark consists of 120 tasks collected from domain experts and a framework for validating and timing LM-synthesized solution code, which is compared to reference implementations from popular open-source packages.

In addition, we develop a baseline LM agent, AlgoTuner, and evaluate its performance across a suite of frontier models. AlgoTuner achieves an average 1.58x speedup against reference solvers, including methods from packages such as SciPy, scikit-learn and CVXPY. However, we find that current models fail to discover algorithmic innovations, instead preferring surface-level optimizations.

We hope that AlgoTune catalyzes the development of LM agents exhibiting creative problem solving beyond state-of-the-art human performance.