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

TL;DR

This paper introduces AlgoTune, a benchmark and framework for testing language models' ability to generate efficient algorithms for complex scientific and mathematical problems, demonstrating modest speedups but limited innovation.

Contribution

It presents AlgoTune, a new benchmark with 154 tasks and a baseline LM agent, AlgoTuner, to evaluate and improve language models' algorithmic code generation capabilities.

Findings

AlgoTuner achieves 1.72x speedup over reference solvers.

Current models tend to optimize surface-level features rather than discover new algorithms.

The benchmark encourages development of more creative and effective algorithmic solutions.

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 (Jimenez et al., 2024) and mathematics (Glazer et al., 2024). 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 154 coding 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 uses a simple, budgeted loop that edits code, compiles and runs it, profiles performance, verifies correctness on tests, and selects the fastest valid version. AlgoTuner achieves an average 1.72x speedup against our reference solvers, which use libraries such as SciPy, sk-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.