Open Source AI Project


OpenBioMed is a groundbreaking open-source project by Tsinghua University's Institute for AI Industry Research (AIR) in collaboration with Shuimu Biosciences.


OpenBioMed, initiated by Tsinghua University’s Institute for AI Industry Research (AIR) and Shuimu Biosciences, is a pioneering open-source endeavor that unveils the world’s first commercially viable, multimodal biopharmaceutical model known as BioMedGPT-10B. This model, featuring tens of billions of parameters, sets a new benchmark in the realm of biopharmaceutical research by delivering unparalleled state-of-the-art performance in cross-modal question-answering tasks. It adeptly handles queries involving natural language, molecules, and proteins, demonstrating expertise on par with human specialists in the biopharmaceutical field. BioMedGPT-10B’s excellence is particularly transformative for critical areas such as drug target discovery, lead compound design and optimization, and protein engineering. The foundation of OpenBioMed lies in its ability to merge vast knowledge spanning amino acids, molecules, and protein data into a cohesive framework. This integration not only fosters a profound comprehension of biological coding mechanisms but also significantly propels advancements in life sciences research and applications.

Further enriching the OpenBioMed project is DrugFM, a foundational model tailored for small molecule drugs. Developed collaboratively by the teams from Tsinghua University’s AIR and the Beijing Institute for General Artificial Intelligence (BIGAI), DrugFM represents a significant leap forward in pharmaceutical research. It cleverly combines the UniMAP small molecule drug pre-training model with the MolFM multimodal molecular foundational model, setting new records in cross-modal retrieval tasks. DrugFM is meticulously designed to grasp the fundamental organization principles and data representation of small molecule drugs, thereby boosting the efficiency of downstream processes such as drug screening, design, and optimization. By incorporating cross-modal attention mechanisms, DrugFM seamlessly bridges molecular structures with corresponding semantic texts, enhancing the comprehension of molecular interactions and fostering innovation in drug development. Collectively, OpenBioMed and its components, including BioMedGPT-10B and DrugFM, stand as monumental contributions to the biopharmaceutical industry, driving both research and practical applications towards new frontiers.

Relevant Navigation

No comments

No comments...