Param 1, a 2.9 billion parameter multilingual LLM, was provided by BharatGen, the government-backed AI effort, as part of its goal to create open source LLMs for Indian academics and developers.
The recently released LLM, known as the “BharatGen Param 1 Indic Scale,” is a pre-trained base model that was completely created from the ground up. It has an astounding 25% of Indic data, which is far more than the 0.01% that is usually utilized in models such as Meta’s Llama.
“Pre-training is an enormous undertaking and often an insurmountable barrier for many. That’s why we’ve taken on this challenge—to provide a robust foundation that you can easily fine-tune for your specific applications,” BharatGen said in a statement.
With AIKosha, developers may now refine the model to create a variety of applications, from India-specific copilots and knowledge systems to Indic chatbots.“With our 2.9 billion parameter base model, we are unlocking new possibilities for innovation and growth across the nation. We hope this sovereign LLM model checkpoint serves as a foundation for India-specific solutions, enabling developers to fine-tune and shape the next generation of AI applications for Bharat,” Prof Ganesh Ramakrishnan, head of BharatGen, told AIM.
In addition to the LLM, the team introduced 20 new speech models (in 19 different Indian languages) with the goal of promoting voice-first interfaces and speech-based innovation for Indian users on AIKosha, the Government of India’s MeitY AI innovation repository.
Under A2TTS-v0.5: Speaker Adaptive TTS, this comprises nine models. These enable developers to produce speech in Marathi, Bengali, Hindi, Gujarati, Tamil, Kannada, Punjabi, Telugu, and Malayalam that mimics the voice of a given speaker.
For Marathi, Tamil, Hindi, Telugu, and Bengali, there are five high-fidelity text-to-speech models available under Speaker-Conditioned TTS (pflow).
Then, under Voicebox TTS Models, there are five more models that are flexible voice synthesis for Bengali, Telugu, Tamil, Marathi, and Hindi.
There is a significant lack of high-quality, publicly accessible speech models for Indian languages, according to BharatGen, which claims that these models were created from the ground up using data that was directly gathered for five Indian languages.
These models now reside in India’s official AI repository, AIKosha, which was introduced by Union Minister Ashwini Vaishnaw. The repository seeks to foster cooperative innovation and centralize India’s AI resources.
“These foundational models are engineered to supercharge India’s AI research and innovation ecosystem,” BharatGen noted, inviting the community to “build an AI that genuinely speaks to, and for, India.”
Along with Ramakrishnan, the contributors of the model include Kundeshwar Pundalik, Durga S, Prateek Chanda, Vedant Goswami, Atul Kumar Singh, Saral Sureka, Panditi Bhagawan, Ajay Nagpal, Smita Gautam, Pankaj Singh, Rishi Bal, and Prof Rohit Saluja.
In a previous interview with AIM, Ramakrishnan stated that BharatGen functions with the distinct goal of “GenAI for Bharat, by Bharat,” in contrast to commercial organizations. Using cost-effective computing and luring top talent from IIT grads, the project will cost less than ₹235 crores, or over $27 million.
IIT Bombay, IIT Kanpur, IIT Mandi, IIT Madras, IIIT Hyderabad, and IIM Indore are all part of the BharatGen collaboration.
BharatGen is a crucial component of Vaishnaw’s prior prediction that India would develop its own fundamental AI models in 7-8 months. “Yes, we are very much on track. The Minister has been briefed, and we are aligned with the timeline,” Ramakrishnan had confirmed earlier. Param-1 is a clear sign of that roadmap.
“Our goal is not just to build AI models but to provide resources that startups and system integrators can leverage,” said Ramakrishnan.
As part of the initiative to build indigenous AI capabilities, MeitY also chose Sarvam AI under the IndiaAI Mission to develop India’s sovereign LLM last month. Work on the 70-billion parameter multimodal AI model that supports both English and Indian languages has already started, according to the team’s proposal.
Also read: Viksit Workforce for a Viksit Bharat
Do Follow: The Mainstream formerly known as CIO News LinkedIn Account | The Mainstream formerly known as CIO News Facebook | The Mainstream formerly known as CIO News Youtube | The Mainstream formerly known as CIO News Twitter |The Mainstream formerly known as CIO News Whatsapp Channel | The Mainstream formerly known as CIO News Instagram
About us:
The Mainstream formerly known as CIO News is a premier platform dedicated to delivering latest news, updates, and insights from the tech industry. With its strong foundation of intellectual property and thought leadership, the platform is well-positioned to stay ahead of the curve and lead conversations about how technology shapes our world. From its early days as CIO News to its rebranding as The Mainstream on November 28, 2024, it has been expanding its global reach, targeting key markets in the Middle East & Africa, ASEAN, the USA, and the UK. The Mainstream is a vision to put technology at the center of every conversation, inspiring professionals and organizations to embrace the future of tech.
