Sarvam to build India’s 1st AI large language model under AI Mission
Science and Technology
4 min read
The Government of India has selected Bengaluru-based startup Sarvam AI to develop India’s first foundational large language model (LLM) under the India AI Mission. Sarvam AI was selected from among 67 applications received by the Government of India for developing the country’s first large language model.
The Indian government's effort to develop indigenous LLM comes in the background of the success of the Chinese startup DeepSeek.
Sarvam AI to develop LLM in Six Months
The government has given Sarvam AI six months to develop the LLM model. The LLM model will be built in India using local infrastructure and engineers. It will use the vast data of various Indian languages and cultures.
The LLM developed by Sarvam will handle voice-based tasks in Indian languages and excel in advanced reasoning.
The model developed by Sarvam AI will have 70 billion parameters and incorporate numerous innovations in programming as well as engineering.
The model is expected to be as good as Openai’s Chat GPT-3 and GPT-4, Meta’s Llama models, etc.
Government Support to Sarvam AI
To support Sarvam, the Government of India will provide 4,000 high-end GPUS (Graphics Processing Units) for six months.
These GPUs will be critical in building the indigenous LLM. Yotta Data Services, Tata Communications, and E2E Networks will make the GPUs available to Sarvam AI for this purpose.
India AI Mission
The government of India launched the IndiaAI Mission on 7th March 2024 with an allocation of Rs 10,3000 crore for five years (2024-2029).
Aim and objectives
- Develop indigenous AI capability to offer solutions in Indian languages and contexts.
- High-end common computing facility in the country with 18,693 Graphics Processing Units (GPUs).
- Development of an indigenous GPU within the next three to five years.
- Develop indigenous semiconductor manufacturing capacity in the country.
- Provide open access to AI infrastructure in the country to startups, researchers, and students at a nominal cost.
About LLM
The large language model (LLM) is a type of artificial intelligence (AI) programme that is capable of understanding and generating natural languages and other types of content.
The LLM is built on machine learning that uses a vast amount of data to recognise and interpret human language or other types of complex data.
The LLM can infer from context, generate coherent and contextually relevant responses, translate to other languages, summarise text, answer questions, etc.
