What question did this study set out to answer?

This research aims to provide a substantial dataset for training multi-modal language-vision models.

November 3, 2021Open Access

LAION-400M: Open Dataset of CLIP-Filtered 400 Million Image-Text Pairs

View Full Paper

Ask AI

Bookmark

Authors

LAION-400M releases a large dataset of image-text pairs, enhancing model training and performance.

This research aims to provide a substantial dataset for training multi-modal language-vision models.
Created LAION-400M dataset consisting of 400 million image-text pairs filtered through CLIP.
Included CLIP embeddings and kNN indices for enhanced similarity search capabilities.
Established the availability of a large-scale dataset facilitating model training.
Enabled efficient similarity search through the inclusion of kNN indices.

Schuhmann et al. (2021) studied this question.

View Full Paper

Ask AI

Bookmark