Datasets
12 resources
Hubs to search first, then the specific corpora that open models are actually trained on. Check the licence before you use any of them — a dataset being downloadable is not the same as it being usable for what you have in mind.
Hugging Face Datasets
The largest open dataset hub
Hundreds of thousands of datasets with a common loading API, previews in the browser, and licence metadata on most cards.
Kaggle Datasets
Community-contributed datasets
Strongest for tabular and competition data, and usually comes with public notebooks showing what others did with it.
Google Dataset Search
A search engine for datasets across the web
Indexes dataset metadata wherever it is published rather than hosting anything itself.
UCI Machine Learning Repository
The long-running academic dataset archive
Small, clean, well-documented classics. Still where a lot of teaching examples come from.
OpenML
Open datasets with shared benchmarks and results
Pairs each dataset with runs and scores from other people, so you can see what performance is reasonable.