OBELICS
OBELICS is a curated collection of 141M English web documents where text and images appear together in the order they had on the original page. Most image-text datasets reduce each image to a single caption, but OBELICS keeps the full document, so a model sees images in the context of the paragraphs around them.
In total it holds 353M images and 115 billion tokens of text. This interleaved structure is what made it suitable for training the IDEFICS family of multimodal models, which read documents rather than isolated image-caption pairs.
The dataset is openly accessible on the Hugging Face Hub under CC BY 4.0.
Related sources
Conceptual Captions (CC3M / CC12M)
A cleaned image alt-text dataset for automatic image captioning, produced by Google Research. Alt-text pulled from the web is filtered and generalised (hypernymed), replacing specific names with broader categories. Available in 3.3M (CC3M) and 12M (CC12M) variants.
DataComp
A benchmark and dataset collection for training multimodal models. It provides a shared pool of image-text candidates and an evaluation framework so teams can test different data-curation strategies against a common yardstick.
LAION-5B
An image-text dataset of over 5.8B examples, built by filtering Common Crawl with a CLIP model that scores how well an image matches a caption. Includes 2.32B English pairs, 2.26B multilingual pairs, and 1.27B not tied to any particular language. Provides URLs, not the images themselves.
PD12M (Public Domain 12M)
An image-text dataset built only from materials marked with a Public Domain Mark or released under Creative Commons Zero (CC0). 12.4M image-caption pairs, with a 3.3M subset called PD3M, sized to match the Conceptual Captions datasets while staying copyright-clean for commercial use.