VeraCruz PT-BR: Portuguese Language Corpus with Regional Origin Labels

190 million Portuguese language samples are organized by regional origin based on URL metadata. The dataset, created by bastao, categorizes content into primary groups for Portugal (PT) and Brazil (BR). It was last updated on July 21, 2025.

Use Cases

• Train language models for Brazilian Portuguese based on the BR-categorized samples.
• Fine-tune models for European Portuguese using the PT-categorized samples.
• Analyze linguistic differences between Brazilian and European Portuguese based on the regional origin labels.
• Benchmark model performance across Portuguese-speaking regions based on the categorized data.

Strengths

• 190 million samples provide a substantial volume of text for training.
• Regional origin categorization (PT/BR) based on URL metadata offers a distinguishing feature for analysis.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

huggingface

Collection Method

Likely gathered from web content, with regional origin inferred from URL metadata.

Freshness

Last updated 2025-07-21 11:53:53; freshness should be verified.

Geography

Portuguese-speaking regions, primarily Portugal and Brazil.