Abstract page for arXiv paper 2611.03727: Auditing a Model-Hub Benchmark: Data Quality, Training Budgets, and Side Information in HuggingBench