Release marks a significant breakthrough in foundational AI model research and its applications for agriculture in China.
China's Nanjing Agricultural University (NAU) has introduced Sinong, the country's first open-source, vertical large language model (LLM) dedicated to the general agricultural sector.
The model is trained on a vast, structured agricultural dataset, covering specialized data from fields such as animal science, agricultural economics and management, agricultural resources and environment, horticulture, smart agriculture, veterinary medicine, plant protection, and crop breeding.
The model, named after the ancient Chinese officials overseeing agriculture and finance, integrates content from nearly 9,000 books, over 240,000 academic papers, approximately 20,000 policy documents and standards, and extensive web-based knowledge.
To tackle common challenges in specialized LLMs such as hallucination and knowledge obsolescence, which often end up with generating incorrect information, the team implemented comprehensive technical solutions. Beyond conventional instruction fine-tuning, the training process incorporated multi-dimensional data, including chain-of-thought and contextual references, significantly enhancing the model's comprehension and generative capabilities for professional agricultural knowledge, according to the report.
Sinong is now fully open-sourced on platforms like ModelScope and GitHub. This open-source strategy aims to reduce barriers to AI application in agriculture, empowering research institutions, enterprises, and developers to build upon Sinong for secondary development and innovation, thereby fostering a collaborative ecosystem for smart agriculture solutions, the NAU was quoted as saying
