TurboEvolve: Towards Fast and Robust LLM-Driven Program Evolution

arXiv cs.AI / 4/22/2026

💬 OpinionIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper introduces TurboEvolve, a multi-island evolutionary framework aimed at making LLM-driven program evolution more sample-efficient and robust within fixed evaluation budgets.
It uses “verbalized Sampling,” prompting the LLM to generate K diverse candidates along with explicit self-assigned sampling weights, combined with an online scheduler that dynamically adjusts K based on stagnation.
TurboEvolve further enhances search by using “seed-pool injection,” which clusters existing solutions and distributes them across islands with controlled perturbations and elitist preservation to balance exploration and refinement.
Experiments on multiple program-optimization benchmarks show TurboEvolve delivers stronger performance under lower budgets and improves best-known solutions on several tasks.

Abstract

LLM-driven program evolution can discover high-quality programs, but its cost and run-to-run variance hinder reliable progress. We propose TurboEvolve, a multi-island evolutionary framework that improves sample efficiency and robustness under fixed evaluation budgets. Inspired by the multiple-offspring strategy in evolutionary algorithms, TurboEvolve introduces verbalized Sampling, prompting the LLM to emit K diverse candidates with explicit self-assigned sampling weights, and an online scheduler that adapts K to expand exploration under stagnation and reduce overhead during steady progress. To exploit existing solution pools, we further propose "seed-pool injection," which clusters seeds and assigns them across islands with controlled perturbations and elitist preservation to balance diversity and refinement. Across multiple program-optimization benchmarks, TurboEvolve consistently achieves stronger performance at lower budgets and improves best-known solutions on several tasks.