VLSP Legal Dataset: Public Test Split for Vietnamese Legal NLP

A Vietnamese legal text dataset from the VLSP legal evaluation benchmark, designed to assess language models' understanding and reasoning capabilities. The dataset contains three distinct evaluation tasks targeting different aspects of legal reasoning and comprehension. It was created by VLSP2025-LegalSML and last updated on July 15, 2025.

Use Cases

• Benchmarking model performance on Vietnamese legal text comprehension based on the described evaluation tasks.
• Evaluating language models' legal reasoning capabilities based on the three distinct task subsets.
• Training or fine-tuning models for specialized legal NLP applications in Vietnamese.

Strengths

• Designed for a specific, structured evaluation benchmark (VLSP legal evaluation benchmark).
• Contains three distinct subsets targeting different evaluation methodologies for legal reasoning.
• Focuses on Vietnamese legal texts, a specialized domain and language combination.

Limitations

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

Provenance

Source

VLSP2025-LegalSML

Freshness

Last updated 2025-07-15 14:41:59; freshness should be verified.