WithinUsAI released a dataset on 2026-01-04 designed for academic-grade fine-tuning of LLMs. It contains 25,000 examples focused on sports rules, sports science, and quantitative sports analytics. The task mix includes 7,000 fact-check/verification items and 10,000 self-contained quantitative reasoning items.
Use Cases
- Fine-tuning LLMs for sports fact verification based on the 7,000 verification items
- Training models on quantitative sports analytics based on the 10,000 quantitative reasoning items
- Developing sports science knowledge models based on the described academic-grade content
- Creating specialized sports rule chatbots based on the dataset's structured task mix
Strengths
- Contains 25,000 examples for LLM fine-tuning
- Includes a fixed mix of 7,000 verification and 10,000 quantitative reasoning tasks
- Designed with a 'Tiny-Recursive-Model-friendly structure'
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
- WithinUsAI
- Freshness
- Last updated 2026-01-04 02:26:58