FPT and NVIDIA expand Nemotron-Personas-Vietnam dataset to 1 million records

FPT and NVIDIA have announced an expansion of the Nemotron-Personas-Vietnam dataset, increasing its scale tenfold from 100,000 to 1 million synthetic user profiles. This open dataset consists of AI-generated personas designed to help developers and researchers build AI models, agents, and applications that better understand Vietnamese language, culture, and demographics.

The synthetic profiles are constructed using official statistical data from Vietnam, including the 2024 Population and Housing Census and the 2024 Vietnam Household Living Standards Survey. Each profile contains 21 data fields, covering six portrait attributes and 15 contextual attributes such as age, gender, education, occupation, location, skills, and interests. The dataset also includes 13,000 unique full-name combinations and 20 occupational groups.

By utilizing NVIDIA’s Nemotron-Personas methodology and open-weight large language models, the project ensures high realism without containing any personally identifiable information (PII). FPT provided the source data, computing infrastructure, and cultural expertise, while NVIDIA supplied the AI methodologies and tools for profile generation.