India is now ready with BharatGPT: India’s first Large Language Model (Generative AI)

• 08 Dec 2023
• TM Team
• 0 Comments

It’s a time to rejoice in the success of having India’s Large Language Model-based solution-BharatGPT. The solution is a combined effort of CoRover.ai, an AI startup company and the support system of I-HUB Anubhuti dedicated to developing data-driven cognitive computing solutions.

CoRover.ai, the creator of BharatGPT is funded by iHub Anubhuti IIITD Foundation, the Technology Innovation Hub of IIIT Delhi, supported by the Department of Science and Technology, Govt of India.

Mr Saurabh Kumar Chaubey, the CEO – iHUB Anubhuti – IIITD Foundation, when contacted on their engagement with BharatGPT, he said “I am incredibly delighted to see what kind of impact BharatGPT is going to create across the board. We at IIIT Delhi and iHub Anubhuti, the National Innovation Hub for Cognitive Computing & Social Sensing, take pride in the support system we have created to support start-ups, innovators, and academic stakeholders in developing data-driven cognitive computing solutions.”

According to Dr Pushpendra Singh, Professor CSE IIITD and Project Director of iHub Anubhuti IIITD Foundation, “In the fast-paced world of business, time is of the essence. BharatGPT, being a light-weight solution, ensures that implementation is swift and efficient. This not only saves valuable time for organizations but also contributes to a seamless integration process.”

In identifying the completeness of BharatGPT, Dr Vikram Goyal, Professor CSE IIITD and Project Director of iHub Anubhuti IIITD Foundation said “A Conversational AI platform also requires other components, such as user interface; a management system; integration with other systems and data sources; voice and video capabilities, to create Virtual Assistants. With all these inbuilt platform centric features already in it, BharatGPT is way ahead of its time.”

One of the key advantages of LLM capabilities is the heightened focus on data security. BharatGPT, offers a responsible and accurate approach, ensuring that sensitive information is handled with the utmost care. The fact that data remains in India is a game-changer. BharatGPT aligns perfectly with the country's vision of "Make AI in India."

BharatGPT, powered by CoRover's Conversational AI platform, efficiently utilizes computing resources through a multi-layered NLP approach, encompassing NLU, NLG, Deep Learning with Generative AI (unsupervised), supervised learning, AIML, and context-based autosuggestion for diverse user query handling.

·       This implies that the burden on the GPUs can be reduced and the whole system can be more resource efficient.

·       Furthermore, BharatGPT is efficient in word embeddings i.e., mapping of a discrete — categorical — variable to a vector of continuous numbers.

·       In the context of neural networks, embeddings are low-dimensional, learned continuous vector representations of discrete variables.

This makes machine learning faster in the case of generative AI and much less memory and compute intensive.

Creating and training large language models demands substantial computational resources, involving vast data and high-performance computing infrastructure. Challenges include data quality, biases, resource efficiency, domain-specific adaptation, continual learning, and security. The execution brilliance can be seen in BharatGPT, which is India’s answer to existing Generative AI/Large Language Models.