Laion

LAIONProjectsTeamBlogNotesPressAboutFAQDonationsPrivacy PolicyDataset RequestsImpressumOpen MenuClose MenuLAIONLarge-scale Artificial Intelligence Open NetworkTRULY OPEN AI. 100% NON-PROFIT. 100% FREE.LAION, as a non-profit organization, provides datasets, tools and models to liberate machine learning research. By doing so, we encourage open public education and a more environment-friendly use of resources by reusing existing datasets and models.Re-LAION 5B release (30.08.2024)LAION-400MAn open dataset containing 400 million English image-text pairs.LAION-5BA dataset consisting of 5.85 billion multilingual CLIP-filtered image-text pairs.Clip H/14The largest CLIP (Contrastive Language-Image Pre-training) vision transformer model.LAION-AestheticsA subset of LAION-5B filtered by a model trained to score aesthetically pleasing images. --- LAIONProjectsTeamBlogNotesPressAboutFAQDonationsPrivacy PolicyDataset RequestsImpressumOpen MenuClose MenuABOUTWe are a non-profit organization with members from all over the world, aiming to make large-scale machine learning models, datasets and related code available to the general public. OUR BELIEF We believe that machine learning research and its applications have the potential to have huge positive impacts on our world and therefore should be democratized. OUR PRINCIPAL GOALS Releasing open datasets, code and machine learning models. We want to teach the basics of large-scale ML research and data management. By making models, datasets and code reusable without the need to train from scratch all the time, we want to promote an efficient use of energy and computing ressources to face the challenges of climate change. FUNDING Funded by donations and public research grants, our aim is to open all cornerstone results from such an important field as large-scale machine learning to all interested communities. --- LAIONProjectsTeamBlogNotesPressAboutFAQDonationsPrivacy PolicyDataset RequestsImpressumOpen MenuClose MenuPROJECTSDATASETSLAION-400Mimage/textStatus: ReleasedFormerly known as crawling@home (C@H), an openly accessible 400M image-text-pair dataset.LAION5Bimage/textStatus: ReleasedA dataset consisting of 5.85 billion CLIP-filtered image-text pairs, featuring several nearest neighbor indices, an improved web-interface for exploration and subset generation, and detection scores for watermark, NSFW, and toxic content detection.Laion-cocoimage/textStatus: Released600M captions generated using BLIP from Laion2B-en.Laion translatedimage/textStatus: Released3B translated samples from Laion5B.Clip H/14image/textStatus: ReleasedThe largest open source clip.LAION5B High-Resimage/textStatus: ReleasedA subset of the LAION5B database, with high resolution images over 1024x1024, containing 170 million samples.LAION Aestheticsimage/textStatus: ReleasedA subset of LAION5B that has been estimated by a model trained on top of clip embeddings to contain only aestheticly pleasing images.LAION-3D3d/image/textStatus: StartedAn effort to create a large-scale dataset consisting of 3D models and descriptor pairs.Audio Datasettext/audioStatus: StartedAn audio dataset for training CLAP and other models, containing a raw and processed dataset, the latter containing .flac files with captions, labels, and other metadata.Watermark Detectionimage/textContrastiveStatus: ReleasedA repository containing datasets to train a watermark classifier.MODELSOpenclipimage/textContrastiveStatus: releasedAn open source implementation of OpenAI's CLIP (Contrastive Language-Image Pre-training).DALLE-2 Prior/Decoderimage/textGenerativeStatus: StartedAn implementation of DALL-E 2, OpenAI's text-to-image synthesis neural network, in Pytorch.ClipCapimage/textGenerativeStatus: ReleasedGenerate text from embedding, using pretrained encoder and language models.CLAPaudio/textContrastiveStatus: StartedA Contrastive Language-Audio Pretraining model, like CLIP, for audio.Video CLIPvideo/textContrastiveStatus: PlanningA contrastive language pretraining model for videos.Multilingual-CLIPimage/textContrastiveStatus: In progressAn implementation of OpenAI's CLIP text encoders for any language.NSFW Detectionimage/textContrastiveStatus: ReleasedA detector for not safe for work content within images using CLIP.Electric Sheepimage/text/audio/videoContrastive/GenerativeStatus: StartedTrain Contrastive and generative models on all modalities.TOOLSimg2datasetimage/textStatus: ReleasedA tool which allows a user to turn large sets of image urls to an image dataset. Can download, resize and package 100M urls in 20 hours on one machine.Clip Retrievalimage/textStatus: ReleasedAllows a user to easily compute clip embeddings and build a clip retrieval system with them. 100M text+image embeddings can be processed in 20 hours using a RTX 3080.Crawlingathome-gpu-hcloudimage/textStatus: ReleasedGPU controlled Hetzner Cloud workers swarm for Crawling@Home project.Clip Benchmarkimage/textStatus: In ProgressEvaluating CLIP-like models on a standard set of datasets on different tasks such as zero-shot classification and zero-shot retrieval.PAPERSLAION-400Mimage/textStatus: PublishedLAION-5Bimage/textStatus: started --- LAIONProjectsTeamBlogNotesPressAboutFAQDonationsPrivacy PolicyDataset RequestsImpressumOpen MenuClose MenuTEAMChristoph SchuhmannOrganizational Lead / FounderMaster in Physics & Computer Science. Producer of Schools of Trust.Jenia JitsevScientific Lead / FounderSenior researcher. Leads Scalable Learning and Multi-Purpose AI (SLAMPAI) Lab at Juelich Supercomputing Center (JSC). PhD in Computer Science, with background in Neuroscience & Machine Learning.Richard VencuEngineering Lead / FounderAI, automation, and electronics engineer with 28 years of experience in the industry.Romain BeaumontOpen source contributor / FounderI like to apply scale and deep learning to build AI apps and models.Robert KaczmarczykCommunity / Medical / Operational Lead / FounderTechnical University of Munich. 3 years experience in epidemiological research.Theo CoombesFounding MemberProgrammer with a great passion for big data and machine learning.Mehdi ChertiFounding Member, senior core researcherResearcher on Machine Learning and Deep Generative Models. Open multi-modal foundation models and datasets, scaling laws. PhD in Computer Science.Irina RishMember, senior researcher.Professor, Computer Science and Operations Research, Université de Montréal.Strong interest in Scaling Laws for MLAran KomatzusakiMember. PhD student of Machine Learning at GaTechInterest in RL, representation learning and generative modeling in generalRobin RombachMember. Stable Diffusion TrainerAI researcher with a focus on deep generative models. Author of VQGAN, Latent Diffusion, Stable Diffusion.Katherine CrowsonMemberAI researcher working on generative models. Generative artistMarianna NezhurinaMember, core researcher multi-modal learning. PhD student JSC/TuebingenCoder and ML researcher with strong interest in open multi-modal foundation models and datasets, scaling laws.Ludwig SchmidtMember, senior lead researcher. Professor in computer science at Stanford University.Research on datasets, reliable generalization, strong reasoning, evaluation and large models.Rio YokotaMember, senior researcher. Professor at Tokyo Institute of Technology, RIKEN. Large-scale machine learning on supercomputers.Research in intersect of high performance computing, linear algebra, and machine learning. Strong interest in open foundation models.Lucia Cipolina-KunMember. PhD Student at the University of Bristol.AI researcher specialized on diffusion models, reinforcement learning and mathematics. Researching on art restoration and preservation of cultural heritage.Dominic RampasTechnische Hochschule IngolstadtAI researcher, passionate about generative models as means to enhance human creative processes. Author of Paella & Würstchen.Pablo PerniasMachine Learning Research en Disney ParksAI researcher, passionate about generative models as means to enhance human creative processes. Author of Paella & Würstchen.Huu NguyenHead of safety policy. Founder and CEO of ontocord.aiComputer scientist and lawyer with over 15 years of experience and advocate for the human rights of equal access to scientific advancement and education.Björn PlüsterUniversity of Hamburg, DiscoResearchAI researcher with a focus on multilingual LLMs and specialized finetuning.Jörg FrankeMember, researcher. PhD student, U Freiburg/ELLIS Institute TuebingenOpen foundation models & datasets, autoML and optimizationAarush KattaFounding MemberProgrammer with lots of enthusiasm for AI & ML.Jan EbertFounding MemberSoftware engineer and researcher, responsible for scaling up deep learning. Helmholtz AI.