IndiaAI Mission Released Indigenous Foundation Models EKA-1 by Soket and Dhīr 2B & Dhīr 5B by ZenteiQ Under the IndiaAI Innovation Centre Pillar

Key Highlights:

• The IndiaAI Mission released indigenous Foundation Models EKA-1 by Soket Labs Technology and Research Private Limited and Dhīr 2B & Dhīr 5B by ZenteiQ AiTech Innovations.
• These models were unveiled under the IndiaAI Innovation Centre pillar of the IndiaAI Mission by the Ministry of Electronics and Information Technology (MeitY).
• EKA-1 is an open, sparse hybrid language model covering all 22 Scheduled Indian languages and English, designed for math, code, and reasoning.
• Dhīr 2B & Dhīr 5B are open-source, secure, on-premise multilingual assistants supporting English and 10 Indian languages, capable of offline deployment.
• The initiative strengthens India's technological and strategic autonomy in Artificial Intelligence, fostering home-grown AI technology.

Detailed Insights:

• The release marks a significant milestone in India's progress towards creating indigenous AI technology and capabilities.
• MeitY Secretary Shri S. Krishnan highlighted that these models enhance productivity across government, industry, research, and education.
• EKA-1 is a 26.3 billion parameter model trained on 7.5 trillion tokens, tailored for India's banking, financial services, and cybersecurity sectors.
• Dhīr models enable document-grounded automation and knowledge retrieval for critical sectors like defence, government, and education, especially in low-connectivity areas.
• Their lightweight architecture allows on-device deployment on mobile phones, ensuring data privacy and accessibility without cloud dependence.
• The IndiaAI Mission, approved on March 7, 2024, aims to democratize technology and position India as a global leader in AI.
• The mission operates across seven pillars, including IndiaAI Compute Capacity, IndiaAI Datasets Platform (AIKosh), and IndiaAI Startup Financing.

Key Concepts Involved:

• IndiaAI Mission: A government initiative approved in March 2024 to establish a robust and inclusive AI ecosystem in India.
• Foundation Models: Large AI models trained on vast datasets that can be adapted for a wide range of downstream tasks.
• Large Language Models (LLMs): A type of foundation model specifically designed to understand and generate human language.
• AIKosh: The IndiaAI Datasets Platform, intended to make AI models accessible to researchers, students, and innovators.