Evaluating In-Context Learning in Large Language Models for Molecular Property Regression.

Journal: Journal of computational chemistry
Published Date: Jan 15, 2026

Abstract

Large language models (LLMs) demonstrate strong performance in natural language tasks, but their capacity for genuine in-context learning (ICL) in scientific regression remains unclear. We systematically assessed seven LLMs on molecular property prediction using a controlled framework of 56 transformed tasks that isolate shortcut learning and are designed to induce functional out-of-distribution (OOD) behavior. LLMs performed nearly perfectly on raw molecular weight prediction via shortcut cues but deteriorated under nonlinear transformations, whereas machine learning (ML) baselines showed greater robustness, yielding a performance crossover. Meta-analysis revealed that distributional descriptors and structure-activity landscape indices (SALI) predict task favorability, providing a framework for selecting between LLM- and ML-based approaches in chemistry.

Authors

  • Chan Young Joe
    Department of Applied Chemistry, University of Seoul, Seoul, Republic of Korea.
  • Kyungwoo Song
    Department of Statistics and Data Science, Yonsei University, 50, Yonsei-ro, Seodaemun-gu, 03722, Seoul, Republic of Korea; Department of Applied Statistics, Yonsei University, 50, Yonsei-ro, Seodaemun-gu, 03722, Seoul, Republic of Korea. Electronic address: [email protected].
  • Rakwoo Chang
    Department of Applied Chemistry, University of Seoul, Seoul, Republic of Korea.

Keywords

No keywords available for this article.

External Resources

Popular Topics
Recent Journals