Adaptive Acquisition Selection for Bayesian Optimization with Large Language Models
Giang Ngo, Dat Phan-Trong, Dang Nguyen, Sunil Gupta & Svetha Venkatesh · 2026
Read Adaptive Acquisition Selection for Bayesian Optimization with Large Language Models by Giang Ngo, Dat Phan-Trong, Dang Nguyen, Sunil Gupta & Svetha Venkatesh (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-07904.txt.
Description
Bayesian tuning hinges on picking the right acquisition function at each step, yet no single rule wins everywhere and the best choice shifts during a run. Giang Ngo and four co-authors propose LMABO, which casts a pretrained language model as a zero-shot online strategist. It reads a structured summary of the optimization state and selects from a portfolio of acquisition functions.
Unlike adaptive methods that look only at past function values, the language model weighs budget remaining, surrogate characteristics, and progress as one. Across fifty benchmark problems, LMABO beats strong static, adaptive, and other language-model baselines. The authors argue the advantage comes from synthesizing the full state into a coherent plan, a case for language models as orchestrators rather than mere predictors.
Catalog record
| CREATOR | Giang Ngo, Dat Phan-Trong, Dang Nguyen, Sunil Gupta & Svetha Venkatesh |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 72 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-07904