Nemotron-Personas-Finance-Japan: A Synthetic Dataset Capturing Japan's Demographics and Financial Context
What happened
Elith and NVIDIA have introduced Nemotron-Personas-Finance-Japan, a synthetic dataset containing 9 million records that mirror Japan's demographic and financial landscape. The dataset was created using official government statistics and advanced synthetic data generation tools. It was developed with input from major Japanese financial institutions to ensure its relevance for financial AI applications. The dataset is available for both commercial and non-commercial use under an open license.
Key details
- 9 million synthetic personas included, each with 50 data fields
- Built using NVIDIA's NeMo Data Designer and NeMo SDG-PGMs frameworks
- Incorporates feedback from Sompo Japan, Mizuho Financial Group, and other industry leaders
- Covers over 1,600 occupation categories and nearly 950,000 unique names
- Released under the CC BY 4.0 license for free commercial use
Why it matters
Japanese banks, insurers, and technology developers can now train and test AI systems for finance without handling real customer data, reducing privacy and compliance risks. The dataset enables these organizations to simulate a wide range of realistic financial scenarios, supporting the development of customer service agents and other financial AI tools. This release provides a resource for any team building Japanese-language AI products that require detailed, representative population data.