LABO: LLM-Accelerated Bayesian Optimization through Broad Exploration and Selective Experimentation

TL;DR

LABO is a novel Bayesian optimization framework that efficiently combines large language models and real experiments, enabling broad exploration and selective testing to accelerate scientific discovery.

Contribution

The paper introduces LABO, a new method that leverages LLMs for exploration and real experiments for precision within a unified Bayesian optimization loop.

Findings

LABO outperforms existing methods under the same experimental budget.

Theoretical analysis provides a regret bound demonstrating efficiency.

Empirical results across scientific tasks validate LABO's effectiveness.

Abstract

The high cost and data scarcity in scientific exploration have motivated the use of large language models (LLMs) as knowledge-driven components in Bayesian optimization (BO). However, existing approaches typically embed LLMs directly into the sampling or surrogate modeling pipeline, without fully leveraging their significantly lower evaluation cost compared to real-world experiments. To address this limitation, we propose LLM-Accelerated Bayesian Optimization (LABO), a framework that combines LLM predictions with experimental observations within a single BO loop. LABO employs a gating criterion to dynamically balance the reliance on LLM predictions versus actual experiments. By leveraging inexpensive LLM evaluations to broadly explore the search space and reserving costly real experiments only for regions with high uncertainty, LABO achieves more sample-efficient optimization. We provide a theoretical analysis with a cumulative regret bound that formalizes this efficiency gain. Empirical results across diverse scientific tasks demonstrate that LABO consistently outperforms existing methods under identical experimental budgets. Our results suggest that LABO offers a practical and theoretically grounded approach for integrating LLMs into scientific discovery workflows.