Mol-LLM: Multimodal Generalist Molecular LLM with Improved Graph Utilization
Chanhui Lee, Hanbum Ko, Yuheon Song, Yongjun Jeong, Rodrigo Hormazabal, Sehui Han, Kyunghoon Bae, Sungbin Lim, Sungwoong Kim · 2025
Read Mol-LLM: Multimodal Generalist Molecular LLM with Improved Graph Utilization by Chanhui Lee, Hanbum Ko, Yuheon Song, Yongjun Jeong, Rodrigo Hormazabal, Sehui Han, Kyunghoon Bae, Sungbin Lim, Sungwoong Kim (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-02810.txt.
Description
Chanhui Lee and eight co-authors teach a language model to read molecules as graphs as well as text, in a 2025 paper. Models trained on letter-string forms of molecules can predict reactions and properties, but a molecule is really a network of atoms, and plain next-token training ignores that structure.
Their model, Mol-LLM, adds a structural pretraining stage that forces the network to use graph information before instruction tuning across many chemistry tasks. The aim is one generalist assistant for molecular work rather than a shelf of narrow tools. Better graph use should sharpen predictions where shape decides behavior, from drug candidates to catalysts, and make the model’s answers more trustworthy across tasks.
Catalog record
| CREATOR | Chanhui Lee, Hanbum Ko, Yuheon Song, Yongjun Jeong, Rodrigo Hormazabal, Sehui Han, Kyunghoon Bae, Sungbin Lim, Sungwoong Kim |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 90 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-02810