Micro LLM aksara Outperforms GPT-4 on Agriculture Queries
Cropin's Krishna Kumar on Addressing Sustainability Challenges Using Generative AI
Technological advancements are revolutionizing the agriculture sector, empowering farmers with data-driven insights to make informed decisions. Cropin has provided more than 7 million farmers access to predictive intelligence data for over 200 million acres of farmland. Cropin works across the agricultural value chain, including packaged goods, farming companies, inputs or seed manufacturers, international organizations, development agencies, and governments.
Cropin's AI lab produces innovations such as aksara - an open-source micro LLM - to address challenges in agriculture logistics, operations and warehousing. aksara is trained on the datasets derived from nine crops in five different countries. Cropin's aksara uses Mistral 7B, a seven billion parameter fundamental model that can answer more than 5,000 questions with a 40% greater efficiency than OpenAI's GPT-4.
"We want our model to be used by researchers, academicians and even the farmers globally to address the sustainability goals and mitigate the adverse impact of climate change on agriculture," said Krishna Kumar, co-founder and CEO of Cropin. "There is no better way to attain this goal than using generative AI based on open-source fundamental models. Proprietary LLMs are biased and thus neglect the Global South."
In this video interview with Information Security Media Group, Kumar also discussed:
- The need for micro LLMs to address industry-specific challenges;
- Idea behind Cropin Cloud and aksara and their operating models;
- How guardrails can be placed to develop AI responsibly.
Kumar is responsible for promoting the development and adoption of digital technologies in the agriculture sector and ensuring Cropin's continued investments in building world-class AI capabilities. His leadership experience of nearly 20 years includes companies such as General Electric.
