English

LLM Program Optimization via Retrieval Augmented Search

Machine Learning 2025-02-03 v1

Abstract

With the advent of large language models (LLMs), there has been a great deal of interest in applying them to solve difficult programming tasks. Recent work has demonstrated their potential at program optimization, a key challenge in programming languages research. We propose a blackbox adaptation method called Retrieval Augmented Search (RAS) that performs beam search over candidate optimizations; at each step, it retrieves in-context examples from a given training dataset of slow-fast program pairs to guide the LLM. Critically, we find that performing contextual retrieval based on an LLM-generated natural language description significantly outperforms retrieval based on the source code. In addition, we propose a method called AEGIS for improving interpretability by decomposing training examples into "atomic edits" that are significantly more incremental in nature. We show that RAS performs 1.8×\times better than prior state-of-the-art blackbox adaptation strategies, and that AEGIS performs 1.37×\times better while performing significantly smaller edits.

Keywords

Cite

@article{arxiv.2501.18916,
  title  = {LLM Program Optimization via Retrieval Augmented Search},
  author = {Sagnik Anupam and Alexander Shypula and Osbert Bastani},
  journal= {arXiv preprint arXiv:2501.18916},
  year   = {2025}
}
R2 v1 2026-06-28T21:27:03.086Z