Skip to content

Dataset · nvidia

Synthetic personas of Brazil

nvidia/Nemotron-Personas-Brazil

A large Portuguese-language dataset of synthetic personas aligned with real Brazilian distributions.

What's inside

Nemotron-Personas-Brazil is an open dataset (CC BY 4.0) of Portuguese synthetic personas. The personas are based on real demographic, geographic distributions, and personality traits of Brazil to reflect population diversity. Size ranges from 1M to 10M records; data is intended for text generation.

What you can do with it

  • Training a Portuguese text-generation model
  • LoRA fine-tuning for Brazilian personas
  • Verifying demographic representation
Tasks
Text generation
Languages
Portuguese
Size
1M–10M rows
Downloads this month
907
Likes
107

How to load it

from datasets import load_dataset

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