Che Boludo Bench: Pragmatic Alignment Benchmark for Rioplatense Spanish

A benchmark dataset for evaluating pragmatic alignment in language models, focusing on the Rioplatense Spanish dialect. The dataset was created by somosnlp-hackathon-2026 and was last updated on June 11, 2026. It is designed to address the tendency of models to interpret language too literally and over-moderate colloquial expressions from specific linguistic communities.

Use Cases

• Benchmarking language model performance on Rioplatense Spanish based on the described focus on dialectal interpretation.
• Training models to better understand cultural and pragmatic nuances in Spanish based on the described limitation of literal processing.
• Developing moderation filters that are more sensitive to regional colloquialisms based on the described issue of over-moderation.
• Researching cross-dialectal alignment in Spanish language models based on the dataset's stated purpose.

Strengths

• Focuses on a specific, culturally rich dialect (Rioplatense Spanish) not commonly benchmarked.
• Addresses a documented limitation in current language models regarding pragmatic and cultural interpretation.
• Last updated on 2026-06-11, suggesting recent maintenance.

Limitations

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

Provenance

Source

huggingface

Freshness

Last updated 2026-06-11 21:18:21; freshness should be verified.

Geography

Likely focused on the Rioplatense region (e.g., Argentina, Uruguay).