---
id: tensorflow-datasets
version: "4.9.10"
license: Apache 2.0
license_treatment: permissive
maintenance: active
---

# tensorflow-datasets — tensorflow/datasets is a library of datasets ready to use with TensorFlow.

License: permissive · Maintenance: active · Downloads: 1.8M/mo

## What it is and what it does

TensorFlow Datasets is a library that provides standardized access to many public datasets, automatically handling download and preparation into tf.data.Dataset objects ready for training pipelines. It wraps datasets from various sources and exposes them through a simple API, letting you load data like MNIST or other benchmarks with a single function call, then chain standard TensorFlow operations like shuffle, batch, and prefetch.

The library emphasizes simplicity for standard use cases, reproducibility (all users get the same examples in the same order), and performance by following TensorFlow best practices. It has 18 runtime dependencies including numpy, pyarrow, and tensorflow-metadata, and requires Python 3.10 or later. The package is actively maintained and carries no known security vulnerabilities.

Use it for:

- Load benchmark datasets like MNIST or CIFAR for quick prototyping and model evaluation.
- Build reproducible input pipelines for training by ensuring deterministic dataset ordering across runs.
- Access a curated catalog of public datasets without manually downloading or preprocessing files.
- Integrate datasets into tf.data pipelines with standard operations like batching and prefetching.
- Prepare datasets for distributed training by leveraging TFDS's performance-optimized loading.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Provides access to many public datasets as tf.data.Datasets, handling download, preparation, and loading for TensorFlow workflows.

Yes. The package is actively maintained, has no known vulnerabilities, and low install friction. It solves a real problem—standardized dataset access for TensorFlow workflows—with a permissive license. Install it if you work with TensorFlow and need quick access to public datasets; skip it only if you manage datasets entirely through custom pipelines.

## Install

pip install tensorflow-datasets
uv add tensorflow-datasets
poetry add tensorflow-datasets

## Installing tensorflow-datasets

Before you install: Low install friction with a pure-Python wheel. Active maintenance (last commit 2026-07-29) and 18 runtime dependencies including numpy, pyarrow, and tensorflow-metadata. Requires Python 3.10 or later.

License in practice: Apache 2.0 permissive license. Users are responsible for determining their own permission to use each underlying dataset; the library itself is freely usable and modifiable.

Quickstart:

pip install tensorflow-datasets
import tensorflow_datasets as tfds
ds = tfds.load('mnist', split='train', as_supervised=True, shuffle_files=True)
ds = ds.shuffle(1000).batch(128).prefetch(10)

Requires Python 3.10 or later and TensorFlow installed separately.

Verify before relying:

- Whether all advertised datasets in the catalog are currently available and maintained.
- Performance characteristics and download speeds for large datasets.
- Compatibility with specific TensorFlow versions beyond the Python requirement.

## Package facts

- License: Apache 2.0 (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 1.8M/month (top 5,000 on PyPI)
- Known vulnerabilities: none known

## Tags

tensorflow datasets loader, public datasets for machine learning, tf.data.Dataset ready-to-use, download and prepare datasets, machine learning dataset catalog, tensorflow data pipeline, dataset management for tensorflow, tensorflow, datasets, machine-learning

[View on SkillFed](https://skillfed.io/packages/tensorflow-datasets) · [View on PyPI](https://pypi.org/project/tensorflow-datasets/)