BoxComm: Boxing Commentary Generation Benchmark

BoxComm-Dataset is the official data release for BoxComm, a benchmark for category-aware boxing commentary generation and narration-rhythm evaluation. The dataset is authored by gouba2333 and was last updated on 2026-05-26. It is intended for training, analysis, and reproducible research.

Use Cases

• Training models for category-aware boxing commentary generation based on the benchmark's defined tasks.
• Evaluating narration rhythm and fluency in generated sports commentary.
• Reproducing research results from the associated paper and project page.

Strengths

• Serves as the official data release for a published benchmark (BoxComm).
• Associated with a research paper (arXiv:2604.04419) and a public GitHub repository.
• Last updated on 2026-05-26, 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.
• The description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

gouba2333

Freshness

Last updated 2026-05-26 04:41:30