Downloading Datasets#

Download a complete snapshot or selected files from a dataset repository. Pin the revision used by an experiment.

1mega snapshot OWNER/DATASET --revision v1.0 --local-dir ./dataset2mega download OWNER/DATASET README.md --revision v1.0 --local-dir ./dataset

For a private or gated dataset, sign in with an account that has access and a token with repo:read. Compatible Hugging Face clients can target MEGA with HF_ENDPOINT; see Hugging Face Compatibility.

For a supported file format, Datasets can load a pinned revision directly:

1from datasets import load_dataset2 3dataset = load_dataset(4    "OWNER/DATASET",5    data_files="data.jsonl",6    split="train",7    revision="v1.0",8    token="YOUR_MEGA_TOKEN",9)10print(dataset)

Use Data Studio when you need a browser preview or SQL exploration rather than a local Datasets object.