Indian genAI startup KissanAI released an experimental research preview of their AI model Dhenu2 for climate resilient and sustainable agriculture practices model.
“This is an experimental release for experts and researchers to provide feedback so that we can work on improvements in subsequent releases,” said Pratik Desai, founder of Kissan AI.
Desai believes that the impact of climate on agriculture is a widely discussed topic, yet much of the conversation and research remains confined to expert and academic circles. The aim is to embed this knowledge into language models that can be integrated into chatbots, assistants, and copilot applications, bringing valuable insights directly to farmers and growers.
The experimental release includes access to model weights and a demo space, allowing experts to explore and test its capabilities. These resources are publicly available on Hugging Face, offering an open platform for feedback and improvement.
By embedding expert agricultural knowledge into AI-driven platforms, Dhenu2 has the potential to revolutionise how farmers adapt to climate challenges. The project’s mission is clear: to democratise knowledge and empower farmers globally with tools that enhance sustainability and resilience in agricultural practices.
Dhenu2 for Farmers
A few months ago, KissanAI had introduced Dhenu2, an advanced agricultural AI platform designed to tackle challenges in the farming sector. Initially targeting India, the platform aimed to expand globally, offering tailored solutions for farmers, policymakers, and agri-businesses.
Powered by the Llama 3 architecture, Dhenu2 is available in three variants , namely, 8B, 3B, and 1B models. Trained on 1.5 million real and synthetic instructions, it spans over 4,000 agricultural categories and topics. This extensive knowledge base is informed by insights gathered from conversations with more than 100,000 farmers, ensuring decision-making rooted in real-world scenarios.
The platform also introduced the Dhenu Developer Platform, enabling developers to create custom agricultural solutions using Dhenu2. Built on six years of curated agricultural data and enhanced by synthetic data pipelines, the open-source nature of the platform provides flexibility for developers and organisations alike.
KissanAI was founded by Desai in 2023, with the aim to address the challenges faced by Indian farmers when it came to region-specific knowledge and digital literacy.
