Skip to content

Dataset · nvidia

NVIDIA synthetic personas for the Japanese financial sector

nvidia/Nemotron-Personas-Finance-Japan

1 to 10 million generated Japanese user profiles considering real demographics for financial AI.

What's inside

This is an extension of NVIDIA's open Nemotron-Personas-Japan dataset under the CC BY 4.0 license, containing 1 to 10 million Japanese-language records (text and images specified in metadata). The profiles are synthetically generated based on real demographic, geographic, and psychological distributions of the Japanese population specifically for financial tasks without personal data leakage risks.

What you can do with it

  • Simulating user behavior and testing financial services in Japan
  • Generating synthetic dialogues and data for SFT without using real personal data
  • Training localized language models for Japanese fintech
Tasks
Text generation
Languages
Japanese
Size
1M–10M rows
Downloads this month
4.4K
Likes
25

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("nvidia/Nemotron-Personas-Finance-Japan")