hugging-face-evaluation

To update Hugging Face model cards with evaluation metrics, this skill automates reading README tables, detecting table formats like rows or transposed layouts, extracting numeric scores, and generating model-index YAML to push structured metadata directly to the model card.

Yes, you can import external benchmarks into Hugging Face model-index metadata. This skill integrates scores from external sources like Artificial Analysis, merging them with existing README evaluation results before generating and pushing the YAML payload.

Yes, this skill supports submitting pull requests for Hugging Face model card updates. You can push model-index YAML updates via PR for models you own or others, or apply direct updates to models you manage.

The best way to automate extraction of numeric scores from markdown tables is using this skill, which detects row, column, and transposed table formats in READMEs, extracts scores, and safeguards against duplicates during model-index YAML generation.

Yes, there are safeguards against duplicate evaluation metrics when generating model-index YAML. This skill implements validation and duplicate detection during the extraction and integration of benchmark results into Hugging Face model cards.