Currently, LLM Eval Tracker only indexes GitHub repositories. Integrating with the Hugging Face Hub would surface evaluation datasets and metrics that aren't on GitHub, significantly expanding coverage. This requires building a new data source adapter that queries the HF Hub API, maps results to our metadata schema, and deduplicates entries across sources.
Currently, LLM Eval Tracker only indexes GitHub repositories. Integrating with the Hugging Face Hub would surface evaluation datasets and metrics that aren't on GitHub, significantly expanding coverage. This requires building a new data source adapter that queries the HF Hub API, maps results to our metadata schema, and deduplicates entries across sources.