Study audits HuggingBench model-hub benchmark for data quality and training budget issues
A new arXiv paper examines HuggingBench, a benchmark built from model-hub data combining interaction logs, knowledge graphs, and documentation, finding that valid files do not guarantee data quality.
Read the full story at arXiv cs.LG ↗
Timeline · 1 report
- 2026-11-04 05:00 · arXiv cs.LG
Auditing a Model-Hub Benchmark: Data Quality, Training Budgets, and Side Information in HuggingBench