Nvidia has unveiled a new artificial intelligence model specifically designed for the Hindi language, named Nemotron-4-Mini-Hindi-4B, during a significant visit by Chief Executive Officer Jensen Huang.
The model, featuring 4 billion parameters, aims to cater to the burgeoning market for AI technologies in India, where the demand for localised solutions is rapidly increasing.
The launch coincided with Huang’s engagement with Mukesh Ambani, chairman of Reliance Industries and Asia’s richest man, at a business conference in Mumbai.
Nvidia’s new offering is set to enable various firms to develop their own tailored AI applications.
According to Nvidia, “The model was pruned, distilled, and trained with a combination of real-world Hindi data, synthetic Hindi data, and an equal amount of
English data.”
This comprehensive approach reflects the company’s commitment to creating versatile tools that meet the needs of a linguistically diverse population.
Indian IT services and consulting firm Tech Mahindra is the first to adopt the new Nvidia model, using it to create a custom AI application named Indus 2.0.
This innovative project will focus on Hindi and its numerous dialects, addressing the fact that only about 10% of India’s 1.4 billion people are English speakers.
Nvidia's model comes as part of a broader strategy to leverage India’s rich linguistic landscape. With 22 languages recognised in the Indian constitution, there is significant potential for AI applications in customer service and content translation.
Unlike large language models such as OpenAI's GPT-4, which power ChatGPT, small language models like Nemotron-4-Mini-Hindi-4B are trained on more specific datasets. This makes them cost-effective, appealing particularly to smaller companies with limited resources.
The company's long-standing presence in India includes engineering and design centres, as well as offices in major tech hubs like Bengaluru and Hyderabad.
Analysts note that while global chip manufacturers are increasing investments in India, establishing a competitive semiconductor industry will take time.
Follow us
