Sarvam AI launches Sarvam 1 for Indic languages
The Hindu
Sarvam AI has launched Sarvam 1, a new AI model which was developed from scratch in India geared towards Indic languages.
Bengaluru-based startup Sarvam AI has launched a new large language model, Sarvam 1. The AI model is open-source and has been trained on 11 languages including Bengali, Gujarati, Hindi, Marathi, Malayalam, Kannada, Oriya, Tamil, Telugu, Punjabi and English.
The 2-billion-parameter model was trained on 4 trillion tokens on a custom tokeniser curated by Sarvam on Nvidia H100 Tensor Core GPUs. The company claims that the tokeniser is up to four times more efficient than other AI models which were trained on Indian languages.
The custom training corpus, Sarvam-2T, comprised of 20% datasets in Hindi, English, and programming languages so the AI model can perform multilingual tasks.
To deal with the lack of high-quality training data for Indian languages, Sarvam AI built datasets using synthetic data generation methods.
Besides Nvidia, the AI model also used Yotta’s data centres and AI4Bharat’s technology and language resources.
“The Sarvam 1 model is the first example of an LLM trained from scratch with data, research, and compute being fully in India”, said Dr. Pratyush Kumar, Co-Founder, Sarvam. He added; “We expect it to power a range of use cases including voice and messaging agents. This is the beginning of our mission to build full stack sovereign AI. We are deeply excited to be working together with NVIDIA towards this mission.”
Developers can use the base model, which is available on Hugging Face, to build their own AI applications for Indic language speakers.
AIADMK faces perception challenges and electoral setbacks in Tamil Nadu, struggling to maintain its traditional stronghold and voter base.
K. Ashok appointed Executive Director of BHEL Tiruchi, succeeding S. Prabhakar after 36 years of experience in the company.
Garbage collection resumes in Bengaluru amid contractor disputes over payment delays and new waste segregation rules by BSWML.
Delhi High Court rules universities cannot restrict peaceful expression of dissent, emphasizing the importance of independent thought and discourse.
Mettur Dam's water level is at 82.83 feet, with an inflow of 384 cusecs and discharge of 800 cusecs.
Lebanon expels Iran's ambassador, declaring him persona non grata, while maintaining a Charge d'Affaires at the embassy.
Sunny Joseph accuses Pinarayi Vijayan of targeting Rahul Gandhi as part of a political deal with the BJP.
AIADMK faces perception challenges and electoral setbacks in Tamil Nadu, struggling to maintain its traditional stronghold and voter base.
Sunny Joseph accuses Pinarayi Vijayan of targeting Rahul Gandhi as part of a political deal with the BJP.
Mahesh Dattani and Jonathan Taikina Taylor collaborate play, ‘The Monk & The Warrior’, to be staged at Mumbai’s Experimental Theatre from April 2 to 5
NHAI's rooftop solar plants on three toll plazas generate 521 daily units, saving 1.87 lakh conventional power units annually.
Congress criticises PM Modi's Rajya Sabha statement on West Asia, calling it a "self-praise" prepared text without Opposition clarification
Kerala has 2.71 crore voters for the April 9 Assembly polls, with 2,040 polling stations marked as critical.