HuggingFace Search Datasets

Search HuggingFace Hub for machine learning datasets by keyword. HuggingFace hosts 100,000+ open datasets spanning text, image, audio, tabular, and biological sequence data. Returns dataset IDs, authors, download counts, likes, tags (including format, modality, size category), and descriptions. Use this to find training/evaluation datasets for ML models, benchmark datasets, or domain-specific datasets (e.g., protein sequences, medical images, genomics).