India’s Sarvam AI brings out large language model Vikram, Pg1

Key Highlights:

• Sarvam AI, based in Bengaluru, launched two open-source language models named Vikram at the AI Impact Summit on Wednesday.
• The Vikram models include a 35-billion parameter model and a 105-billion parameter model.
• These models aim to improve performance in Indian languages, addressing the scarcity of training data.
• Sarvam AI received over $50 million in funding from investors like Peak XV and Khosla Ventures.

Detailed Insights:

• The launch of Vikram is considered a significant milestone in AI development in India, aligning with the government's push for indigenous LLMs.
• Training LLMs requires substantial computational resources, involving GPUs and significant investment.
• The Ministry of Electronics and Information Technology (MeitY) is focusing on both inference and training of LLMs in India.
• Sarvam AI utilized subsidized access to GPUs under the IndiaAI Mission’s common compute program.
• The models' key objective is to enhance AI's proficiency in understanding and processing various Indian languages.

Key Concepts Involved:

• Large Language Model (LLM): An AI model trained on vast amounts of text data to generate human-like text.
• Inference: The process of using a trained AI model to make predictions or generate outputs based on new input data.
• GPU (Graphics Processing Unit): A specialized electronic circuit designed to rapidly manipulate and alter memory to accelerate the creation of images in a frame buffer intended for output to a display device.

Working of LLMs