17 Apr 2024
Micro model is cheaper and better energy efficient than large language models.
Bengaluru-based Agriculture intelligence company Cropin launched an open-source microlanguage model, Aksara, to facilitate smart agriculture practices. The model will address farmer problems in the Global South regions, which includes Africa, Latin America, the Caribbean, Asia excluding South Korea and Japan, and Oceania, excluding Australia and New Zealand.
It is based on Mistral’s seven billion parameters foundation generative text model. It offers best practices and ideas for proper seeding, nursery preparation, and water conservation. It covers nine crops, including millets, maize, paddy, wheat, cotton, soybean, barley, sorghum, and sugarcane, cultivated in India, Bangladesh, Sri Lanka, Nepal, and Pakistan. The model will accommodate more languages besides English, more regions, and crops in the next six months.
Founded in 2010, Cropin sells agriculture-intelligent solutions in over 103 countries. The model promotes sustainable agricultural practices and energy efficiency. It can enable underserved farmers to create cost-effective, scalable AI-based smart agriculture solutions.