LARI: Brazilian Portuguese Question-Answer Triplets for Educational Benchmarking

LARI is a native Question and Answer dataset in Brazilian Portuguese containing 464 context-question-answer trios validated by human experts. It was created by jjuliar to address the scarcity of educationally grounded benchmarks for Portuguese, as opposed to automatically translated resources. The dataset was last updated on 2026-04-26.

Use Cases

• Benchmarking Portuguese-language question-answering models based on native, expert-validated trios
• Training language models for educational content in Brazilian Portuguese based on the described context-question-answer structure
• Evaluating the performance of machine translation against native Portuguese QA resources based on the dataset's stated purpose

Strengths

• Contains 464 expert-validated context-question-answer trios
• Built from native Portuguese content, not automatic translations, addressing a specific scarcity

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is known (464), but other metadata like file formats and license are unknown

Provenance

Source

jjuliar on Hugging Face

Collection Method

Built from content originally in Portuguese and validated by human experts

Freshness

Last updated 2026-04-26 21:12:29; freshness should be verified

Geography

Brazilian Portuguese language focus