Ola Introduces कृत्रिम (Krutrim), India’s own multi-lingual AI for 1.4 Bn+ Indians
कृत्रिम (Krutrim) unveiled its base Large Language Model (LLM). With the largest representation of Indian data used for its training powers generative AI applications for all Indian languages. Trained by a team of leading computer scientists, based in Bangalore and San Francisco, this model will also power कृत्रिम (Krutrim’s) conversational AI assistant that understands and speaks multiple Indian languages fluently.
Bhavish Aggarwal, Founder, कृत्रिम (Krutrim) said, “It is our time to realise the dream of Viksit Bharat and lead the AI vision over the next 25 years. AI is going to define the future paradigm of economy and culture. We are extremely excited to launch India’s first complete AI computing stack, कृत्रिम (Krutrim), which understands our unique cultural context, connecting our future to our roots. With an India-first cost structure, कृत्रिम (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.”
Ola कृत्रिम (Krutrim) is a family of Large Language Models, including कृत्रिम (Krutrim) base and कृत्रिम (Krutrim) Pro which will have multimodal, larger knowledge capabilities, and many other technical advancements for inference. Trained on over 2 trillion tokens, कृत्रिम (Krutrim) accomplishes better performance on multiple well known, global, LLM evaluation benchmarks including MMLU, HellaSwag, BBH, PIQA and ARC.
Ola कृत्रिम (Krutrim) model can fluently switch between languages and discuss nuanced topics ranging from poetry in Bengali, to Bollywood movies, to creative masala dosa recipes! कृत्रिम (Krutrim) will be available in beta version from January 2024 as an API for enterprises and developers, seeking to create AI-driven assistants that can converse in multiple Indian languages. कृत्रिम (Krutrim) Pro will be available in Q4, FY 2024.
Ola कृत्रिम’s (Krutrim) superior linguistic skills make it a valuable tool for a wide range of purposes from education to business communications. It incorporates the latest techniques in safe AI to reduce inappropriate responses. The company is also working on AI infrastructure to develop indigenous data centres and eventually, server-computing, edge-computing and super-computers. Production is scheduled for mid 2024 for prototypes and a roll out production roadmap by the end of 2025.
The company is running an early access program starting today until January 2024 with a simple sign up page on the website. Additionally, the Open release is scheduled for January 2024 and open APIs will open to all developers by February 2024. Ola also plans to use कृत्रिम (Krutrim) across its group companies for its sales, service, support and other processes.
Key highlights of कृत्रिम (Krutrim):
– Generative support for 10 Indian languages today and understanding of all 22 Indian languages
– Trained on over 2 trillion tokens of data, with a unique mix of training data for understanding/generating text in Indian languages
– Fine-tuned on a large volume of Indian data to understand the nuances of Indian culture and languages
– Uses a custom tokenizer for understanding different Indian languages and scripts
– Performs better than many open-source LLMs trained with similar amounts of data on a suite of industry standard benchmarks including MMLU, HellaSwag, BBH, PIQA and ARC.
