AI research firm Soket AI Labs has collaborated with Google Cloud to unveil Pragna-1B, an open-source multilingual foundation language model.
Pragna-1B is proficient in Hindi, English, Bengali and Gujarati, developed using data specific to Indian contexts. The collaboration between Soket AI Labs and Google Cloud extends beyond the launch of Pragna-1B.
Plans are reportedly underway to list Soket’s AI Developer Platform on the Google Cloud Marketplace and introduce the Pragna series of models to the Google Vertex AI model registry. This integration aims to provide developers with a streamlined experience for fine-tuning models, leveraging the intuitive interface of Soket’s AI Developer Platform and the high-performance resources of Vertex AI and TPUs.
In a statement, Abhishek Upperwal, founder at Soket AI Labs said, “Our partnership with Google Cloud has been pivotal in the pre-training of the Pragna-1B model. By leveraging Google Cloud’s AI Infrastructure we achieved both efficiency and cost-effectiveness in the development of Pragna-1B. Despite being trained on fewer parameters, Pragna-1B showcases unparalleled efficiency and innovation to achieve comparable performance and effectiveness in language processing tasks to similar category models.”
“Tailored specifically for vernacular languages, Pragna-1B offers balanced language representation and enables faster and more efficient tokenisation suited for organizations seeking optimized operations and enhanced functionality,” Upperwal added.
The collaboration with Google Cloud is expected to facilitate more technical cooperation on foundational work for training large-scale models and curating high-quality datasets for Indian languages.
“We are thrilled to partner with Soket AI Labs to democratize AI innovation in India. Built on Google Cloud, the launch of Pragna -1B marks a pioneering leap in Indian language technology, offering enhanced scalability and efficiency for organisations,” said Bikram Singh Bedi, Vice President and Country MD, Google Cloud India.