Where does the data come from?

We aggregate datasets from Hugging Face, Kaggle, GitHub, UCI ML Repository, and government open data portals.

Do I need an API key?

No. Basic search and browsing works without any API key. Some features (like downloading from Kaggle) require their free API key.

Can I filter by commercial license?

Yes, use the license filter to find datasets with permissive, commercial-friendly licenses like MIT, CC-BY-4.0, or Apache-2.0.

Dataset Browser

Search 400K+ public AI/ML datasets from Hugging Face, Kaggle, and more.

12 datasets

imdb

IMDB movie reviews for sentiment analysis — 50K labeled reviews

Hugging FaceMIT

Text·Text Classification·80 MB

cifar100

CIFAR-100 tiny images dataset — 100 classes, 60K images

KaggleMIT

Image·Image Classification·161 MB

common voice

Mozilla Common Voice — crowdsourced speech dataset in 100+ languages

Hugging FaceCC0

Audio·Speech Recognition·~100 GB

squad

Stanford Question Answering Dataset — 100K+ Q&A pairs

Hugging FaceCC-BY-SA-4.0

Text·Question Answering·~30 MB

celeba

CelebFaces Attributes Dataset — 200K celebrity images with attributes

KaggleResearch Only

Image·Face Recognition·1.5 GB

librispeech

LibriSpeech ASR corpus — 1000 hours of English speech

Hugging FaceCC-BY-4.0

Audio·Speech Recognition·~60 GB

wikipedia

Wikipedia dataset — full Wikipedia articles for pretraining

Hugging FaceCC-BY-SA-3.0

Text·Language Modeling·~20 GB

mnist

MNIST handwritten digits — 70K grayscale images, 10 classes

KaggleCC-BY-SA-3.0

Image·Image Classification·~12 MB

alpar alpaca gpt4

Alpaca-style instruction tuning dataset generated by GPT-4

GitHubMIT

Text·Instruction Tuning·~25 MB

fineweb

FineWeb — curated web dataset for LLM training, 15T tokens

Hugging FaceODC-BY

Text·Language Modeling·~10 TB

open images

Open Images v7 — 9M+ images with bounding boxes and labels

KaggleCC-BY-4.0

Image·Object Detection·~20 GB

xsum

Extreme Summarization dataset — BBC articles with summaries