G42 Launches Hindi LLM NANDA at UAE-India Business Forum

G42, a prominent technology holding group based in the UAE, has unveiled plans to introduce NANDA, a state-of-the-art Hindi Large Language Model (LLM). This upcoming launch demonstrates G42’s commitment to advancing AI capabilities in the Hindi language.

Specifications of NANDA

NANDA is a 13-billion parameter model, developed using approximately 2.13 trillion tokens of language data, with a significant focus on the Hindi language. These vast datasets are integral to the model’s proficiency in understanding and generating text in Hindi.

Collaboration and Technological Backbone

The name "NANDA" is inspired by one of India's most majestic peaks, symbolizing the high aspirations of the project. The development of NANDA was made possible through the collaboration of three key entities: Inception (a subsidiary of G42), Mohamed bin Zayed University of Artificial Intelligence (MBZUAI)—the world’s first graduate research institution dedicated to artificial intelligence—and Cerebras Systems. Training for NANDA was conducted using the Condor Galaxy, one of the world’s most powerful AI supercomputers, built by G42 and Cerebras to advance AI model training and inferencing.

“India has solidified its position as a global technology leader, driven by transformative initiatives like Digital India and Startup India under Prime Minister Narendra Modi’s leadership. As the country stands on the brink of AI-powered growth, G42 is proud to contribute to this journey with the launch of NANDA in support of India's AI ambitions,” says Manu Jain, CEO – G42 India.

“G42 has a strong track record in the development of language and domain-specific LLMs. With NANDA, we are heralding a new era of AI inclusivity, ensuring that the rich heritage and depth of the Hindi language is represented in the digital and AI landscape. NANDA exemplifies G42’s unwavering commitment to excellence and fostering equitable AI,” says Dr. Andrew Jackson, Acting CEO of Inception, a G42 company.

Impact on Hindi AI and Language Communities

The upcoming release of NANDA is poised to revolutionize the AI landscape for over half a billion Hindi speakers. By making cutting-edge generative AI capabilities available to this community, NANDA aims to unlock unprecedented potential for AI-powered applications in the Hindi language.

G42’s Previous Success: JAIS and its Influence

In August 2023, G42 made waves by introducing JAIS, the first open-source Arabic language model, which transformed the Arabic Natural Language Processing (NLP) field. With models ranging from 590 million to 70 billion parameters, JAIS democratized access to generative AI for more than 400 million Arabic speakers globally, setting new benchmarks for AI in linguistics. G42 is now determined to replicate this success with NANDA, focusing on regions and languages that are still underrepresented in the AI landscape.

G42’s Broader Mission with NANDA

NANDA extends G42’s mission to empower India’s scientific, academic, and developer communities. By fostering a thriving Hindi AI ecosystem, G42 seeks to provide broad and equitable access to artificial intelligence, supporting innovation and growth across India’s AI landscape.