SELFIES

SELFIES is a molecular representation format where every string corresponds to a valid molecule, used to evaluate the ability of large language models to generate valid molecular structures.

Papers using SELFIES (3)

How to Make Large Language Models Generate 100% Valid Molecules?2025BioT5: Enriching Cross-modal Integration in Biology with Chemical Knowledge and Natural Language Associations2023 · 37 citesHierarchical Structure Enhances the Convergence and Generalizability of Linear Molecular Representation2024