---
source: https://iinventedlight.com/library/notes/hf-datasets-cloud-storage.md
title: "Move dataset files from cloud storage with fsspec"
license: CC-BY-4.0
license_url: https://creativecommons.org/licenses/by/4.0/
source_license: "Apache-2.0 (huggingface/datasets docs); restated in our own words"
attribution: "Summary by iinventedlight (AI-assisted). Sources: Hugging Face (Datasets docs), Cloud storage · Hugging Face, Apache-2.0 (huggingface/datasets docs); Hugging Face, Stream, Apache-2.0 (huggingface/datasets docs)."
author: "iinventedlight (synthesis of cited sources); text drafted with AI assistance"
date: 2026-11-04
status: published
checked_on: 2026-11-05
as_of: 2026-11-05
confidence: medium
untested: "No code was run; only the fetched Hugging Face documentation pages were read."
category: huggingface-datasets
topic_area: "datasets: storage"
tags: huggingface, datasets, fsspec, cloud-storage
sources:
  - id: S1
    title: "Cloud storage · Hugging Face"
    publisher: Hugging Face (Datasets docs)
    url: https://huggingface.co/docs/datasets/filesystems
    accessed: 2026-11-05
    license: "Apache-2.0 (huggingface/datasets docs)"
  - id: S2
    title: "Stream"
    publisher: Hugging Face
    url: https://huggingface.co/docs/datasets/stream
    accessed: 2026-11-05
    license: "Apache-2.0 (huggingface/datasets docs)"
    license: "Apache-2.0 (huggingface/datasets docs)"
---
# Hugging Face Datasets: cloud storage and streaming

## Short answer
Load from cloud filesystems through fsspec-compatible stores, or stream with `IterableDataset` when you do not want a full local Arrow conversion. [S1][S2]

## Points
- Most cloud providers have an fsspec FileSystem implementation. [S1]
- Non-versioned buckets lack the git-style history Hub repos provide. [S1]
- Streaming avoids waiting for huge local conversions and saves disk. [S2]

## Claims
- id: C1
  claim: Most cloud providers have an fsspec FileSystem.
  source: S1
  evidence: "Most cloud storage providers have a fsspec FileSystem implementation"
- id: C2
  claim: Hub repos provide data versioning via git.
  source: S1
  evidence: "provides data versioning thanks to git"
- id: C3
  claim: Streaming works without downloading the whole dataset.
  source: S2
  evidence: "work with a dataset without downloading it"
- id: C4
  claim: Streaming helps when converted size would exceed disk space.
  source: S2
  evidence: "exceed the amount of available disk space on your computer"

## Sources
- [S1] Cloud storage · Hugging Face. Hugging Face (Datasets docs). https://huggingface.co/docs/datasets/filesystems . License: Apache-2.0 (huggingface/datasets docs). Accessed 2026-11-05.
- [S2] Stream. Hugging Face. https://huggingface.co/docs/datasets/stream . License: Apache-2.0 (huggingface/datasets docs). Accessed 2026-11-05.

License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Summary by iinventedlight (AI-assisted).

Source notice: Hugging Face Datasets documentation, Apache License 2.0 (https://github.com/huggingface/datasets/blob/main/LICENSE).