Krutrim, an AI startup founded by Ola CEO Bhavish Aggarwal, is working on a multilingual large language model (LLM). Once built, the LLM will be able to generate text in 10 Indian languages including Hindi, Marathi, Telugu, Kannada and Odiya. The model is built to understand macronic text that blends Hindi and English.
Agarwal noted that the model is work in progress as he demoed it at an event in Bengaluru. It has been trained on over two trillion tokens using a custom tokenizer so that the LLM can understand nuances in Indian languages.
A Pro version of the AI model could be released next quarter. It will be able to address “complex problem-solving and task execution capabilities.”
(For top technology news of the day, subscribe to our tech newsletter Today’s Cache)
“With an India-first cost structure, Krutrim will have the largest representation of Indian data, enabling us to create novel models beyond LLMs across sectors, making India the most productive, efficient and empowered economy in the world,” Agarwal said in a statement.
The company also said that it is on the path to develop indigenous data centres and eventually move towards “server computing, edge computing and super computers.” The production for this project is scheduled for mid-next year and the roadmap will be rolled out by the end of 2025.
The model has an early access program running until January 2024 where interested users can sign up. The model is scheduled for a release in January and the beta version will be accessible to developers by February next year via APIs.
month
Please support quality journalism.
Please support quality journalism.
Wanda Parisien is a computing expert who navigates the vast landscape of hardware and software. With a focus on computer technology, software development, and industry trends, Wanda delivers informative content, tutorials, and analyses to keep readers updated on the latest in the world of computing.
