Multimodal C4: An Open, Billion-scale Corpus of Images Interleaved with Text
This provides a foundational resource for researchers and developers working on in-context vision and language models, addressing a key data bottleneck in the field.
The authors tackled the lack of publicly available large-scale datasets with interleaved images and text for training multimodal models, and they released Multimodal C4, a corpus with 101.2M documents, 571M images, and 43B tokens, where 88% of images are topically relevant and 80% are well-aligned with sentences.
In-context vision and language models like Flamingo support arbitrarily interleaved sequences of images and text as input. This format not only enables few-shot learning via interleaving independent supervised (image, text) examples, but also, more complex prompts involving interaction between images, e.g., "What do image A and image B have in common?" To support this interface, pretraining occurs over web corpora that similarly contain interleaved images+text. To date, however, large-scale data of this form have not been publicly available. We release Multimodal C4, an augmentation of the popular text-only C4 corpus with images interleaved. We use a linear assignment algorithm to place images into longer bodies of text using CLIP features, a process that we show outperforms alternatives. Multimodal C4 spans everyday topics like cooking, travel, technology, etc. A manual inspection of a random sample of documents shows that a vast majority (88%) of images are topically relevant, and that linear assignment frequently selects individual sentences specifically well-aligned with each image (80%). After filtering NSFW images, ads, etc., the resulting corpus consists of 101.2M documents with 571M images interleaved in 43B English tokens.