Recently, Guangxi University officially launched the self-developed smart agriculture large language model platform—DeepSeek AgrDS_V0 (Agriculture DeepSeek). This innovative achievement was jointly developed by the State Key Laboratory for Conservation and Utilization of Subtropical Agro-Bioresources, the School of Agriculture, and the School of Computer, Electronic and Information, marking a significant step forward for Guangxi University in the field of agricultural artificial intelligence research and application.

The DeepSeek AgrDS_V0 platform is developed based on a localized corpus, integrating research methodologies from biology and agronomy. It consolidates over 2.47 million relevant literature pieces and enhances the model's capabilities in scientific reasoning, topic selection, literature summary, and mathematical reasoning through advanced Chain-of-Thought (CoT) training techniques. Its core functionality lies in providing intelligent Q&A services tailored to biology and agronomy. Currently, the platform supports formula parsing and mathematical derivation in LaTeX format, offering seamless assistance for faculty and students in research activities. The deployment of this platform not only demonstrates Guangxi University’s technological expertise in smart agriculture but also provides a secure and efficient research tool for the entire academic community. Operated entirely on the university’s internal servers, the platform ensures data security, accelerates model response speeds, and serves as a showcase for the institution’s achievements in agricultural AI technology transformation.  

Since September 2022, Guangxi university has been laying the groundwork in the field of AI-driven smart agriculture. The launch of the DeepSeek AgrDS_V0 platform represents a significant breakthrough in smart agriculture and a practical milestone in smart agricultural research, providing robust technical support for the research and innovation of all faculty and students. Moving forward, Guangxi university plans to expand the platform’s functionalities by integrating specialized databases and Retrieval-Augmented Generation (RAG) technologies to enhance its capabilities in genomics, molecular biology, and related fields. Simultaneously, the next-generation 671B-parameter V1 iteration is already under development, aiming to deliver even more precise and user-friendly research services for the university community.