ContextEvolve: Multi-Agent Context Compression for Systems Code Optimization

TL;DR

ContextEvolve introduces a multi-agent framework that enhances systems code optimization by decomposing context into semantic, directional, and experiential components, achieving reinforcement learning-level efficiency without parameter updates.

Contribution

It presents a novel multi-agent approach that enables efficient, training-free optimization in a textual latent space, overcoming API access limitations and improving performance on code optimization benchmarks.

Findings

Outperforms state-of-the-art baselines by 33.3% on ADRS benchmark.

Reduces token consumption by 29.0%.

Demonstrates effective optimization without parameter updates.

Abstract

Large language models are transforming systems research by automating the discovery of performance-critical algorithms for computer systems. Despite plausible codes generated by LLMs, producing solutions that meet the stringent correctness and performance requirements of systems demands iterative optimization. Test-time reinforcement learning offers high search efficiency but requires parameter updates infeasible under API-only access, while existing training-free evolutionary methods suffer from inefficient context utilization and undirected search. We introduce ContextEvolve, a multi-agent framework that achieves RL-level search efficiency under strict parameter-blind constraints by decomposing optimization context into three orthogonal dimensions: a Summarizer Agent condenses semantic state via code-to-language abstraction, a Navigator Agent distills optimization direction from trajectory analysis, and a Sampler Agent curates experience distribution through prioritized exemplar retrieval. This orchestration forms a functional isomorphism with RL-mapping to state representation, policy gradient, and experience replay-enabling principled optimization in a textual latent space. On the ADRS benchmark, ContextEvolve outperforms state-of-the-art baselines by 33.3% while reducing token consumption by 29.0%. Codes for our work are released at https://anonymous.4open.science/r/ContextEvolve-ACC