Skip to content

Dataset · nvidia

Synthetic data for Japanese PII detection

nvidia/Nemotron-PII-Japan

NVIDIA dataset for training and evaluating Japanese personally identifiable information (PII) detection and masking models.

What's inside

A synthetic dataset by NVIDIA ranging from 100k to 1 million examples in Japanese for token-classification tasks. Built using fictional personas, the data covers Japanese names, addresses, identifiers, financial details, organization names, and work documents, enabling model training without using real confidential texts.

What you can do with it

  • Training NER models to detect and anonymize PII in Japanese texts
  • Creating security filters and DLP systems for Japanese corporate document workflows
  • Testing the extraction accuracy of sensitive entities in financial and business documents
Tasks
Token classification
Languages
Japanese
Size
100K–1M rows
Downloads this month
679
Likes
12

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("nvidia/Nemotron-PII-Japan")