AI RESEARCH
Speak-to-Structure: Evaluating LLMs in Open-domain Natural Language-Driven Molecule Generation
arXiv CS.CL
•
ArXi:2412.14642v4 Announce Type: replace Recently, Large Language Models (LLMs) have nstrated great potential in natural language-driven molecule discovery. However, existing datasets and benchmarks for molecule-text alignment are predominantly built on one-to-one mappings, measuring LLMs' ability to retrieve a single, pre-defined answer, rather than their creative potential to generate diverse, yet equally valid, molecular candidates.