Reproduce “From Words to Molecules: A Survey of Large Language Models in Chemistry” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Chang Liao, Yemin Yu, Yu Mei, Ying Wei
This survey systematically categorizes and analyzes the application of Large Language Models (LLMs) in chemistry, detailing how molecular information is encoded, the various pretraining objectives, and emerging application paradigms like LLMs as interactive chatbots and in-context learners. The work from Nanyang Technological University, City University of Hong Kong, and Zhejiang University provides a structured understanding of this rapidly evolving interdisciplinary domain.
Key insights
- LLMs can process chemical information effectively through various encoding and tokenization methods, including data-driven motif-level approaches like BPE.
- The field is progressing from single-domain chemical LLMs to multi-domain and multi-modal models that integrate natural language and richer chemical data.
- LLMs are being leveraged beyond representation learning, demonstrating capabilities as interactive chatbots for chemical queries and in-context learners for new tasks.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.