Skip to content

Dataset · nvidia

Multilingual profiles of Belgian residents from NVIDIA

nvidia/Nemotron-Personas-Belgium

Between 1 and 10 million synthetic personas of Belgian residents in four languages, mapped to real statistical distributions.

What's inside

The dataset contains between 1 and 10 million text descriptions of Belgian resident personas in Dutch, French, German, and English. The profiles are generated using a compound AI approach and rely on real demographic distributions.

What you can do with it

  • Generating synthetic dialogues and role-play scenarios considering local context
  • SFT training of language models for response personalization in Belgian languages
Tasks
Text generation
Languages
Dutch, French, German, English
Size
1M–10M rows
Downloads this month
675
Likes
33

How to load it

from datasets import load_dataset

ds = load_dataset("nvidia/Nemotron-Personas-Belgium")