Authors: Pengfei Liu, Yiming Ren, Jun Tao, Zhixiang Ren

Published on: August 14, 2023

Impact Score: 8.22

Arxiv code: Arxiv:2308.06911

Summary

• What is new: GIT-Mol integrates Graph, Image, and Text information of molecules into one model, unlike prior models that handled less complex data.
• Why this is important: Previous language models couldn’t effectively process the rich, complex structures or images of molecules.
• What the research proposes: Introducing GIT-Mol, a multi-modal large language model that uses GIT-Former architecture to unify different molecular data modalities into one model.
• Results: Achieved a 5%-10% increase in accuracy for property prediction and a 20.2% boost in molecule generation validity.

Technical Details

Technological frameworks used: GIT-Former

Models used: GIT-Mol, a multi-modal large language model

Data used: Graph, Image, and Text information of molecules

Potential Impact

Pharmaceuticals, chemical manufacturing, and AI-driven bioinformatics companies.

Want to implement this idea in a business?

We have generated a startup concept here: MolSight.