Work

GOLLuM: Gaussian Process-Optimized LLMs

Year: 2026

Type: software

Abstract: Code and data to reproduce the experiments in the paper. GOLLuM combines a Gaussian process surrogate with an LLM encoder finetuned through the GP marginal likelihood, enabling uncertainty-calibrated Bayesian optimization from natural language inputs across 23 chemistry and materials benchmarks.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.20625386

Author: Ranković, Bojana

Language: English

Topic: Machine Learning in Materials Science

Subfield: Materials Chemistry

Field: Materials Science

Domain: Physical Sciences

Open Access status: green

OpenAlex ID: W7164137386

Cites:

Cited by: 1

Related to: 10