The STEM-specific large language model is a collaboration between SPREE, CSE, the UNSW AI Institute and National Computational Infrastructure, and has already attracted international funding.
Large language models (LLM) like ChatGTP are revolutionising how researchers can sort through vast amounts of data for varied applications in their work. At SPREE, PhD candidate Tong Xie has collaborated with colleagues at the UNSW School of Computer Science and Engineering (CSE) on developing and launching an LLM that is specifically trained on peer-reviewed, STEM research papers: Darwin. Its powerful algorithm is also trained on chemistry and physics journals, as well as materials science,making it a highly useful new tool for researchers working across these fields.
Xie’s GreenDynamics, with SPREE, CSE, National Computational Infrastructure and the UNSW AI Institute, opens a new chapter in AI-assisted natural science research through the Darwin LLM. Securing access to the publications’ databases for educational and research purposes, the model continues to learn from emerging research. Its capabilities will continue to expand, making it an indispensable tool for researchers worldwide.
Darwin’s natural language understanding, generation capabilities, and scalability open up exciting new avenues to enhance the efficiency and effectiveness of materials development and new applications. GreenDynamics recently completed a successful round of funding for Darwin from INNOANGEL (Hong Kong), and Australia’s Viresent Venture.
Code and further details are available to use on GitHub.
For collaboration queries, please contact Tong Xie: tong.xie@unsw.edu.au