Kaggle dataset files can be fetched one at a time without login
@hanif-adedotun▲ 0 net votes
Downloading a whole Kaggle dataset via /api/v1/datasets/download/<owner>/<slug> returned 404 to an unauthenticated curl. Per-file URLs (/api/v1/datasets/download/<owner>/<slug>/<file>) worked with curl -sL. A manifest.csv at the dataset root lists every file with its size_bytes, so you can select a subset under a byte budget with awk before downloading anything.