NVIDIA Korean Synthetic Dataset Tops Hugging Face Rankings

Santa Clara, CA - NVIDIA (NVDA.O) said Tuesday that its Korean-specific synthetic dataset, "Nemotron-Personas-Korea," has reached the top ranking in the dataset category on the AI development platform Hugging Face.

The dataset consists of 6 million synthetic records designed to reflect South Korea's demographic, geographic, and cultural characteristics. 

The company said the data was constructed from public and private sources, including the Korean Statistical Information Service (KOSIS), the Supreme Court of Korea, the National Health Insurance Service, the Korea Rural Economic Institute, and NAVER Cloud.

NVIDIA said the dataset aligns with statistical distributions across several categories, including name, gender, age, marital status, education level, occupation, and residential area. 

Technical specifications also account for linguistic elements such as the Korean honorific system and regional occupational patterns.

The company said the 6 million records include data points for demographic groups that are often underrepresented in standard datasets, such as the elderly, rural residents, and diverse vocational groups.

On privacy, NVIDIA said Nemotron-Personas-Korea contains no personally identifiable information and was built to comply with South Korea's Personal Information Protection Act.

The dataset is available under an open-source license. NVIDIA said it expects the release to help developers expand data diversity, reduce model bias, and improve response quality in Korean-language AI systems.

The announcement followed "Nemotron Developer Days Seoul 2026," an event the company held to engage researchers and corporate developers on open models and data-driven applications.