Hugging Face Hub launches Evaluation on the Hub powered by AutoTrain

AI Impact Summary

Hugging Face Hub introduces Evaluation on the Hub, a no-code tool powered by AutoTrain that lets you evaluate any model on any dataset with any metric. Results are encoded in model card metadata and a PR is opened on the Hub to surface the evaluation; this enables reproducible benchmarking and centralized visibility via leaderboards and the model-evaluator Space. DistilBERT's model card is used as an example, illustrating how evaluation metadata is surfaced alongside models. This shifts model-quality decision-making toward auditable, code-free evaluation artifacts that underpin deployment and governance decisions.