Ola CEO Bhavish Aggarwal took to the stage in Bengaluru to introduce India’s own AI model called Krutrim. The new AI model was also given a demo during the event where the AI model was able to respond in different Indian languages. The Krutrim AI model can also code in real-time. According to Aggarwal, the new AI model will be capable of understanding Indian context better, compared to AI models trained mostly on Western data.
Read More: Adani Group acquires over 50% stake in news agency IANS: Report
Krutrim has been introduced in two classifications, the base LLM model called Krutrim which has been trained on 2 trillion tokens. The model can understand 22 languages and can generate output in 10 Indian languages. The company claims that the new Krutrim AI outperforms GPT-4 and Llama models in terms of Indic performance.
The company claims that Krutrim AI has been trained on 20 times more indic tokens than any other model. The new model can use voice as input and the model will also respond in voice.
Krutrim team in Bengaluru also mentioned a more powerful model in the works: Krutrim Pro. This specific model will be multi-modal however it will be made available by next quarter.
Grounds up for India
According to the company, the new model has been built from the ground up and the entire stack has been built by Krutrim. Aggarwal during the Q&A session said, “India needs a solution that is thought grounds up for itself. For India to be an AI-first economy we need to build the whole stack at the Indian performance levels, Indian cultural relevance and Indian cost structure.”
He added, “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, Krutrim, which understands our unique cultural context, connecting our future to our roots.
Read More: Google CEO Sundar Pichai opens up on laying off over 12,000 employees; says ‘very difficult but…’
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.”
Krutrim Availability
The new Krutrim AI model website is live and interested customers can register their interest in the AI model. The user can register using their phone number and the company will inform them as soon as the chatbot is made available for them. Aggarwal confirmed that the base LLM will be made available to everyone from next month. The company is also building APIs which will be made available in February.
Key Highlights of Krutrim:
Currently, Krutrim can generate content in 10 Indian languages and can comprehend all 22 Indian languages
It has been trained on a massive dataset of over 2 trillion tokens, with a distinctive blend of data specifically designed for understanding and generating text in Indian languages
To grasp the subtleties of Indian culture and languages, it has been fine-tuned on a substantial amount of Indian data
Krutrim employs a custom tokenizer to interpret various Indian languages and scripts
Read More: India’s Exports Dip 2.83% To $33.9 Bn In Nov; Trade Deficit Narrows To $20.58 Bn: Govt Data
When compared to numerous open-source Language Learning Models (LLMs) trained with similar data volumes, Krutrim outperforms them on a range of industry-standard benchmarks, including MMLU, HellaSwag, BBH, PIQA, and ARC.
