Nvidia Unveils AI Model For Hindi Language

At its AI summit in Mumbai, Nvidia, the global leader in chip manufacturing, declared the release of its latest AI model tailored for the Hindi language, further strengthening its presence in India. The declaration came as Nvidia's CEO, Jensen Huang, met with Mukesh Ambani, Chairman of Reliance Industries, to discuss expanding Nvidia’s computing infrastructure in India. Nvidia aims to increase its capacity in the country by 20 times over the next year.

The newly released AI model, named Nemotron-4-Mini-Hindi-4B, is designed to cater to the Hindi-speaking population in India, which comprises hundreds of millions of people. The model has 4 billion parameters and has been trained using a mix of real-world and synthetic Hindi data, as well as English data, allowing it to support both languages. It is aimed at enterprises developing AI solutions in areas like customer service and content translation, particularly for Hindi speakers.

Industry Adoption And Collaboration

Indian IT services company Tech Mahindra is among the first to adopt Nvidia’s new AI model, using it to develop Indus 2.0, an AI system designed to serve Hindi speakers and support various dialects. The move addresses the need for AI models to focus on regional languages, as English is spoken by only 10 per cent of India’s population.

Nvidia is also collaborating with major Indian technology firms, including Infosys, TCS and Wipro, to train 500,000 developers to build AI applications, focusing upon Nvidia’s long-term investment in fostering local AI talent and supporting the growth of the AI ecosystem in India.

Along with its AI initiatives, Nvidia is working with companies such as Reliance and Ola Electric to leverage its Omniverse technology. This technology is being used to simulate factory operations, offering insights into how AI can revolutionise industries such as manufacturing and electric vehicles in India.