KisanSLM is a compact domain-specific language model for agricultural advisory in India. Fine-tuned on the Kisan Call Center (KCC) dataset, it learns from millions of farmer queries and expert responses on crops, pests, soil, irrigation, and fertilizers. Designed for low-resource environments, KisanSLM can run on edge devices and supports multilingual queries, enabling farmers to receive timely, practical guidance even in low-connectivity rural settings.
KisanSLM – Small Language Model for Agricultural Advisory KisanSLM is a compact, domain-specific language model designed to provide practical agricultural guidance to farmers across India. The model is fine-tuned on the Kisan Call Center (KCC) dataset, which contains millions of farmer queries and expert responses collected by the Ministry of Agriculture & Farmers’ Welfare. This dataset captures real-world farming challenges such as crop diseases, pest management, soil health, irrigation practices, fertilizer usage, and market-related questions. The model learns from the patterns and recommendations present in these expert responses, enabling it to generate clear, actionable advice aligned with the realities faced by smallholder farmers. Because the training data reflects actual farmer interactions, KisanSLM understands the context, terminology, and practical constraints common in Indian agriculture. KisanSLM is designed to operate in low-resource environments. Its lightweight architecture allows it to run on edge devices or local servers, making it suitable for deployment in rural areas where internet connectivity may be unreliable. This enables agricultural advisory systems to function offline or with minimal bandwidth, ensuring farmers can access guidance when they need it most. The model also supports multilingual interaction, reflecting the linguistic diversity present in the KCC dataset. Farmers can ask questions in regional languages and receive responses that are contextually relevant and understandable. By incorporating agro-climatic and regional knowledge embedded in the training data, KisanSLM can generate recommendations that are sensitive to local conditions. By distilling the knowledge of agricultural experts from thousands of real advisory conversations, KisanSLM helps bridge the gap between institutional agricultural expertise and farmers in the field. It can be integrated into mobile applications, helplines, kiosks, or conversational interfaces, enabling farmers to receive timely, reliable advice on crop management and farming practices. KisanSLM aims to make agricultural knowledge more accessible, scalable, and responsive—supporting farmers with an always-available advisory companion designed specifically for the realities of Indian agriculture.
Apache 2.0
Aakash Gupta
Fine-Tuned Model
Gemma C++
Open
AgriKosh
09/03/26 11:50:30
150.77 MB
11 files, 1 directories
Apache 2.0
© 2026 - Copyright AIKosh. All rights reserved.