LAION AI

LAION (Large-scale Artificial Intelligence Open Network) is a non-profit organization aiming to make large-scale machine learning models, datasets, and related code available to the general public. LAION encourages open public education and an environmentlyenvironmentally friendly use of resources by reusing existing datasets and models. LAION is funded by donations and public research grants. The organization was founded in the summer of 2021 by researchers Christoph Schuhmann (organizational lead), Jenia Jitsev (scientific lead), Richard Vencu (engineering lead), Robert Kaczmarczyk (Medical Lead), Theo Coombes, Mehdi Cherti, Aarush Katta, and Jan Ebert. The organization is made up of global members who collaborate remotely on LAION projects.

LAION has developed a number projects that include large-scale AI models, datasets for training AI models, and tools related to image/text systems. Notable LAION projects include the following:

• LAION-400M—openLAION-400M—an datassetopen dataset with 400 million English image-text pairs
• LAION-5B—datsetLAION-5B—a dataset consisting of 5.85 billion multilingual CLIP-filtered image-text pairs
• LAION Aesthetics—a subset of LAION-5B filtered by a model trained to find aesthericallyaesthetically pleasing images

The organization's LAION-5B dataset has been used to train numerous AI models, including Stable Diffusion and Google's Imagen. Stability AI funded the creation of LAION-5B and worked with LAION to develop the aesthteticsaesthetics subset. The initial version of Stable Diffusion was based on LAION-400M, LAION-5B's predecessor. The datasets are known to contain depictions of sex, slurs, and harmful stereotypes. LAION-Aesthetics acts as an attempt to correct these issues. Creating LAION-5B meant scraping large repositories of artwork, including DeviantArt, ArtStation, Pinterest, and Getty Images, and others. Getty Images has begun legal action against Stable Diffusion for using its libraryslibraries to train its models (via the LAION-5B dataset) and infringing copyright. Getty has broughbrought lawsuits against Stable Diffusion in Delaware and London, UK.

LAION (Large-scale Artificial Intelligence Open Network) is a non-profit organization aiming to make large-scale machine learning models, datasets and related code available to the general public. LAION encourages open public education and an environmently friendly use of resources by reusing existing datasets and models. LAION is funded by donations and public research grants. The organization was founded in summer 2021 by German researchers Christoph Schuhmann (organizational lead), Jenia Jitsev (scientific lead), and Richard Vencu (engineering lead), Robert Kaczmarczyk (Medical Lead), Theo Coombes, Mehdi Cherti, Aarush Katta, and Jan Ebert. The organization is made up of global members who collaborate remotely on LAION projects.

LAION (Large-scale Artificial Intelligence Open Network) is a non-profit organization aiming to make large-scale machine learning models, datasets and related code available to the general public. LAION encourages open public education and an environmently friendly use of resources by reusing existing datasets and models. LAION is funded by donations and public research grants. The organization was founded in summer 2021 by German researchers Christoph Schuhmann (organizational lead), Jenia Jitsev (scientific lead), and Richard Vencu (engineering lead).

LAION has developed a number projects that include large-scale AI models, datasets for training AI models, and tools related to image/text systems. Notable LAION projects include:

• LAION-400M—open datasset with 400 million English image-text pairs
• LAION-5B—datset consisting of 5.85 billion multilingual CLIP-filtered image-text pairs
• Clip H/14—a CLIP (Contrastive Language-Image Pre-training) vision transformer model
• LAION Aesthetics—a subset of LAION-5B filtered by a model trained to find aestherically pleasing images

The organization's LAION-5B dataset has been used to train numerous AI models including Stable Diffusion and Google's Imagen. Stability AI funded the creation of LAION-5B and worked with LAION to develop the aesthtetics subset. The initial version of Stable Diffusion was based on LAION-400M, LAION-5B's predecessor. The datasets are known to contain depictions of sex, slurs and harmful stereotypes. LAION-Aesthetics acts as an attempt to correct these issues. Creating LAION-5B meant scraping large repositories of artwork including DeviantArt, ArtStation, Pinterest, Getty Images, and others. Getty Images has begun legal action against Stable Diffusion for using its librarys to train its models (via the LAION-5B dataset) and infringing copyright. Getty has brough lawsuits against Stable Diffusion in Delaware and London, UK.

In March 2023, LAION began a petition calling for an international organization, similar to CERN, to coordinate and progress large-scale AI research safely and transparently.