diff --git a/404/index.html b/404/index.html index 5e89967f..fc82a94a 100644 --- a/404/index.html +++ b/404/index.html @@ -1 +1 @@ -

LAION

404 - Not Found

Go back home
\ No newline at end of file +

LAION

404 - Not Found

Go back home
\ No newline at end of file diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/about.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/about.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/about.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/about.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog-de.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog-de.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog-de.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog-de.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog-de/leo-lm.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog-de/leo-lm.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog-de/leo-lm.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog-de/leo-lm.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/bud-e.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/bud-e.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/bud-e.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/bud-e.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/clara-release.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/clara-release.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/clara-release.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/clara-release.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/coca.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/coca.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/coca.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/coca.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/datacomp.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/datacomp.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/datacomp.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/datacomp.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/falling-walls-2023.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/falling-walls-2023.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/falling-walls-2023.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/falling-walls-2023.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/giant-openclip.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/giant-openclip.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/giant-openclip.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/giant-openclip.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/h14_clip_retrieval.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/h14_clip_retrieval.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/h14_clip_retrieval.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/h14_clip_retrieval.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/laion-400-open-dataset.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/laion-400-open-dataset.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/laion-400-open-dataset.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/laion-400-open-dataset.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/laion-5b.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/laion-5b.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/laion-5b.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/laion-5b.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/laion-aesthetics.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/laion-aesthetics.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/laion-aesthetics.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/laion-aesthetics.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/laion-coco.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/laion-coco.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/laion-coco.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/laion-coco.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/laion-pop.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/laion-pop.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/laion-pop.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/laion-pop.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/laion-stable-horde.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/laion-stable-horde.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/laion-stable-horde.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/laion-stable-horde.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/laion-translated.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/laion-translated.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/laion-translated.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/laion-translated.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/large-openclip.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/large-openclip.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/large-openclip.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/large-openclip.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/leo-lm.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/leo-lm.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/leo-lm.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/leo-lm.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/objaverse-xl.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/objaverse-xl.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/objaverse-xl.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/objaverse-xl.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/oig-dataset.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/oig-dataset.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/oig-dataset.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/oig-dataset.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/open-empathic.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/open-empathic.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/open-empathic.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/open-empathic.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/open-flamingo-v2.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/open-flamingo-v2.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/open-flamingo-v2.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/open-flamingo-v2.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/open-flamingo.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/open-flamingo.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/open-flamingo.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/open-flamingo.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/open-lm.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/open-lm.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/open-lm.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/open-lm.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/paella.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/paella.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/paella.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/paella.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/petition.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/petition.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/petition.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/petition.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/relaion-5b.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/relaion-5b.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/relaion-5b.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/relaion-5b.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/strategic-game-dataset.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/strategic-game-dataset.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/strategic-game-dataset.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/strategic-game-dataset.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/transparent-ai.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/transparent-ai.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/transparent-ai.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/transparent-ai.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/video2dataset.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/video2dataset.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/video2dataset.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/video2dataset.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/blog/visit_bench.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/blog/visit_bench.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/blog/visit_bench.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/blog/visit_bench.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/donations.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/donations.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/donations.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/donations.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/impressum.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/impressum.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/impressum.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/impressum.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/notes.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/notes.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/notes.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/notes.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/notes/cpretrain.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/notes/cpretrain.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/notes/cpretrain.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/notes/cpretrain.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/notes/general-gpt.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/notes/general-gpt.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/notes/general-gpt.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/notes/general-gpt.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/notes/laion-debate.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/notes/laion-debate.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/notes/laion-debate.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/notes/laion-debate.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/notes/laion-maintenance.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/notes/laion-maintenance.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/notes/laion-maintenance.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/notes/laion-maintenance.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/notes/letter-to-the-eu-parliament.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/notes/letter-to-the-eu-parliament.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/notes/letter-to-the-eu-parliament.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/notes/letter-to-the-eu-parliament.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/notes/open-gpt-4-o.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/notes/open-gpt-4-o.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/notes/open-gpt-4-o.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/notes/open-gpt-4-o.json diff --git a/_next/data/ExXEuuC6X1gZlFqBuFE3l/notes/realfake.json b/_next/data/a90J7cHEDIdnRAFKcu9sY/notes/realfake.json similarity index 100% rename from _next/data/ExXEuuC6X1gZlFqBuFE3l/notes/realfake.json rename to _next/data/a90J7cHEDIdnRAFKcu9sY/notes/realfake.json diff --git a/_next/static/ExXEuuC6X1gZlFqBuFE3l/_ssgManifest.js b/_next/static/ExXEuuC6X1gZlFqBuFE3l/_ssgManifest.js deleted file mode 100644 index 3c139919..00000000 --- a/_next/static/ExXEuuC6X1gZlFqBuFE3l/_ssgManifest.js +++ /dev/null @@ -1 +0,0 @@ -self.__SSG_MANIFEST=new Set(["\u002Fabout","\u002Fblog","\u002Fdonations","\u002Fimpressum","\u002Fnotes","\u002Fblog-de","\u002Fblog\u002F[slug]","\u002Fnotes\u002F[slug]","\u002Fblog-de\u002F[slug]"]);self.__SSG_MANIFEST_CB&&self.__SSG_MANIFEST_CB() \ No newline at end of file diff --git a/_next/static/ExXEuuC6X1gZlFqBuFE3l/_buildManifest.js b/_next/static/a90J7cHEDIdnRAFKcu9sY/_buildManifest.js similarity index 96% rename from _next/static/ExXEuuC6X1gZlFqBuFE3l/_buildManifest.js rename to _next/static/a90J7cHEDIdnRAFKcu9sY/_buildManifest.js index 04379225..849e4030 100644 --- a/_next/static/ExXEuuC6X1gZlFqBuFE3l/_buildManifest.js +++ b/_next/static/a90J7cHEDIdnRAFKcu9sY/_buildManifest.js @@ -1 +1 @@ -self.__BUILD_MANIFEST=function(s,a,e){return{__rewrites:{beforeFiles:[],afterFiles:[],fallback:[]},"/":["static/css/7b5652f665fe5ccd.css","static/chunks/pages/index-0c98b821f124dca8.js"],"/404":["static/chunks/pages/404-3adee414c9d6d223.js"],"/_error":["static/chunks/pages/_error-0a004b8b8498208d.js"],"/about":[s,"static/chunks/pages/about-99fb9cbc374aaab0.js"],"/blog":[s,a,"static/chunks/pages/blog-21908775d8c17a89.js"],"/blog/[slug]":[s,e,"static/chunks/pages/blog/[slug]-44fac0971625f498.js"],"/blog-de":[s,a,"static/chunks/pages/blog-de-71d7fe712644fb13.js"],"/blog-de/[slug]":[s,e,"static/chunks/pages/blog-de/[slug]-9b92a19caa39d7d8.js"],"/dataset-requests":["static/chunks/pages/dataset-requests-d96d7a4689f00977.js"],"/donations":[s,"static/chunks/pages/donations-a604893cb7771045.js"],"/faq":["static/chunks/pages/faq-b15f9b94b1f8b696.js"],"/impressum":[s,"static/chunks/pages/impressum-21d49f5415a91e29.js"],"/laion-400-open-dataset":["static/chunks/pages/laion-400-open-dataset-bd060d6a2499a8ac.js"],"/laion-5b-a-new-era-of-open-large-scale-multi-modal-datasets":["static/chunks/pages/laion-5b-a-new-era-of-open-large-scale-multi-modal-datasets-9097b5051adddf15.js"],"/notes":[s,a,"static/chunks/pages/notes-14fe44fb8a8ad8c0.js"],"/notes/[slug]":[s,e,"static/chunks/pages/notes/[slug]-2b0138ebd04b8900.js"],"/press":["static/chunks/pages/press-f2724375625fc903.js"],"/privacy-policy":["static/chunks/pages/privacy-policy-dd5753e48a5983de.js"],"/projects":["static/chunks/pages/projects-4cdccd5cc4d4eaf7.js"],"/team":[a,"static/chunks/pages/team-f537bf491f420de7.js"],sortedPages:["/","/404","/_app","/_error","/about","/blog","/blog/[slug]","/blog-de","/blog-de/[slug]","/dataset-requests","/donations","/faq","/impressum","/laion-400-open-dataset","/laion-5b-a-new-era-of-open-large-scale-multi-modal-datasets","/notes","/notes/[slug]","/press","/privacy-policy","/projects","/team"]}}("static/chunks/286-30519d8a3e60551d.js","static/chunks/58-229d4b5119b416a2.js","static/chunks/807-a4eae1dfa8bfbe9f.js"),self.__BUILD_MANIFEST_CB&&self.__BUILD_MANIFEST_CB(); \ No newline at end of file +self.__BUILD_MANIFEST=function(s,a,e){return{__rewrites:{beforeFiles:[],afterFiles:[],fallback:[]},"/":["static/css/7b5652f665fe5ccd.css","static/chunks/pages/index-0c98b821f124dca8.js"],"/404":["static/chunks/pages/404-3adee414c9d6d223.js"],"/_error":["static/chunks/pages/_error-0a004b8b8498208d.js"],"/about":[s,"static/chunks/pages/about-99fb9cbc374aaab0.js"],"/blog":[s,a,"static/chunks/pages/blog-21908775d8c17a89.js"],"/blog/[slug]":[s,e,"static/chunks/pages/blog/[slug]-44fac0971625f498.js"],"/blog-de":[s,a,"static/chunks/pages/blog-de-71d7fe712644fb13.js"],"/blog-de/[slug]":[s,e,"static/chunks/pages/blog-de/[slug]-9b92a19caa39d7d8.js"],"/dataset-requests":["static/chunks/pages/dataset-requests-d96d7a4689f00977.js"],"/donations":[s,"static/chunks/pages/donations-a604893cb7771045.js"],"/faq":["static/chunks/pages/faq-b15f9b94b1f8b696.js"],"/impressum":[s,"static/chunks/pages/impressum-21d49f5415a91e29.js"],"/laion-400-open-dataset":["static/chunks/pages/laion-400-open-dataset-bd060d6a2499a8ac.js"],"/laion-5b-a-new-era-of-open-large-scale-multi-modal-datasets":["static/chunks/pages/laion-5b-a-new-era-of-open-large-scale-multi-modal-datasets-9097b5051adddf15.js"],"/notes":[s,a,"static/chunks/pages/notes-14fe44fb8a8ad8c0.js"],"/notes/[slug]":[s,e,"static/chunks/pages/notes/[slug]-2b0138ebd04b8900.js"],"/press":["static/chunks/pages/press-cb9558c605827176.js"],"/privacy-policy":["static/chunks/pages/privacy-policy-dd5753e48a5983de.js"],"/projects":["static/chunks/pages/projects-4cdccd5cc4d4eaf7.js"],"/team":[a,"static/chunks/pages/team-f537bf491f420de7.js"],sortedPages:["/","/404","/_app","/_error","/about","/blog","/blog/[slug]","/blog-de","/blog-de/[slug]","/dataset-requests","/donations","/faq","/impressum","/laion-400-open-dataset","/laion-5b-a-new-era-of-open-large-scale-multi-modal-datasets","/notes","/notes/[slug]","/press","/privacy-policy","/projects","/team"]}}("static/chunks/286-30519d8a3e60551d.js","static/chunks/58-229d4b5119b416a2.js","static/chunks/807-a4eae1dfa8bfbe9f.js"),self.__BUILD_MANIFEST_CB&&self.__BUILD_MANIFEST_CB(); \ No newline at end of file diff --git a/_next/static/ExXEuuC6X1gZlFqBuFE3l/_middlewareManifest.js b/_next/static/a90J7cHEDIdnRAFKcu9sY/_middlewareManifest.js similarity index 100% rename from _next/static/ExXEuuC6X1gZlFqBuFE3l/_middlewareManifest.js rename to _next/static/a90J7cHEDIdnRAFKcu9sY/_middlewareManifest.js diff --git a/_next/static/a90J7cHEDIdnRAFKcu9sY/_ssgManifest.js b/_next/static/a90J7cHEDIdnRAFKcu9sY/_ssgManifest.js new file mode 100644 index 00000000..8c04320a --- /dev/null +++ b/_next/static/a90J7cHEDIdnRAFKcu9sY/_ssgManifest.js @@ -0,0 +1 @@ +self.__SSG_MANIFEST=new Set(["\u002Fabout","\u002Fblog-de","\u002Fdonations","\u002Fimpressum","\u002Fnotes","\u002Fblog","\u002Fblog\u002F[slug]","\u002Fnotes\u002F[slug]","\u002Fblog-de\u002F[slug]"]);self.__SSG_MANIFEST_CB&&self.__SSG_MANIFEST_CB() \ No newline at end of file diff --git a/_next/static/chunks/pages/press-cb9558c605827176.js b/_next/static/chunks/pages/press-cb9558c605827176.js new file mode 100644 index 00000000..dbca529b --- /dev/null +++ b/_next/static/chunks/pages/press-cb9558c605827176.js @@ -0,0 +1 @@ +(self.webpackChunk_N_E=self.webpackChunk_N_E||[]).push([[313],{8923:function(e,t,n){(window.__NEXT_P=window.__NEXT_P||[]).push(["/press",function(){return n(817)}])},8396:function(e,t,n){"use strict";n.d(t,{Z:function(){return o}});var r=n(5893),i=n(9008),a=n.n(i),s=n(1163);function o(e){var t=(0,s.useRouter)(),n=e.title?e.title+" | LAION":"LAION",i=e.desc?e.desc:"LAION, Large-scale Artificial Intelligence Open Network, is a non-profit organization making machine learning resources available to the general public.",o=e.image?e.image:"/social.png",u=e.alt?e.alt:"The text: LAION. Large-scale Artificial Intelligence Open Network, TRULY OPEN AI. 100% NON-PROFIT. 100% FREE.",l=e.slug?e.slug:t.route;return(0,r.jsxs)(a(),{children:[(0,r.jsx)("title",{children:n}),(0,r.jsx)("meta",{name:"title",content:n}),(0,r.jsx)("meta",{property:"og:title",content:n}),(0,r.jsx)("meta",{name:"twitter:title",content:n}),(0,r.jsx)("meta",{name:"description",content:i}),(0,r.jsx)("meta",{property:"og:description",content:i}),(0,r.jsx)("meta",{name:"twitter:description",content:i}),(0,r.jsx)("meta",{property:"og:image",content:"https://laion.ai"+o}),(0,r.jsx)("meta",{name:"twitter:image",content:"https://laion.ai"+o}),(0,r.jsx)("meta",{name:"twitter:image:alt",content:u}),(0,r.jsx)("meta",{property:"og:type",content:"website"}),(0,r.jsx)("meta",{property:"og:url",content:"https://laion.ai"+l}),(0,r.jsx)("meta",{name:"twitter:url",content:"https://laion.ai"+l}),(0,r.jsx)("meta",{name:"twitter:card",content:"summary_large_image"}),(0,r.jsx)("meta",{name:"viewport",content:"initial-scale=1.0, width=device-width"}),(0,r.jsx)("meta",{name:"theme-color",content:"#1D374E"}),(0,r.jsx)("link",{rel:"icon",type:"image/png",sizes:"32x32",href:"/favicon.png"}),(0,r.jsx)("link",{rel:"icon",href:"/favicon.svg",type:"image/svg+xml"}),(0,r.jsx)("link",{rel:"preload",href:"/fonts/DinishCondensed-Bold.woff",as:"font",type:"font/woff",crossOrigin:"true"}),(0,r.jsx)("link",{rel:"preload",href:"/fonts/DinishCondensed-Bold.woff2",as:"font",type:"font/woff2",crossOrigin:"true"}),(0,r.jsx)("link",{rel:"preload",href:"/fonts/Dinish-Regular.woff",as:"font",type:"font/woff",crossOrigin:"true"}),(0,r.jsx)("link",{rel:"preload",href:"/fonts/Dinish-Regular.woff2",as:"font",type:"font/woff2",crossOrigin:"true"}),(0,r.jsx)("link",{rel:"preload",href:"/fonts/Dinish-Italic.woff",as:"font",type:"font/woff",crossOrigin:"true"}),(0,r.jsx)("link",{rel:"preload",href:"/fonts/Dinish-Italic.woff2",as:"font",type:"font/woff2",crossOrigin:"true"})]})}},817:function(e,t,n){"use strict";n.r(t),n.d(t,{default:function(){return o}});var r=n(5893),i=n(8396),a=JSON.parse('[{"dt":"August 30 2024","url":"https://www.arstechnica.com/tech-policy/2024/08/nonprofit-scrubs-illegal-content-from-controversial-ai-training-dataset"},{"dt":"August 30 2024","url":"https://www.heise.de/en/news/RE-LAION-5B-Image-database-without-abuse-images-9852925.html"},{"dt":"August 30 2024","url":"https://www.techcrunch.com/2024/08/30/the-org-behind-the-data-set-used-to-train-stable-diffusion-claims-it-has-removed-csam/"},{"dt":"August 30 2024","url":"https://www.apnews.com/article/ai-image-generators-child-sexual-abuse-laion-stable-diffusion-2652b0f4245fb28ced1cf74c60a8d9f0"},{"dt":"June 10 2024","url":"https://www.heise.de/news/Reasoning-Fail-Gaengige-LLMs-scheitern-an-kinderleichter-Aufgabe-9755034.html"},{"dt":"February 16 2024","url":"https://www.nature.com/articles/d41586-024-00497-8"},{"dt":"February 15 2024","url":"https://www.techcrunch.com/2024/02/15/this-german-nonprofit-is-building-an-open-voice-assistant-that-anyone-can-use/"},{"dt":"February 14 2024","url":"https://www.heise.de/news/KI-Update-kompakt-ChatGPT-personalisieren-Digitalisierung-IAFM-Bud-E-9628384.html"},{"dt":"February 13 2024","url":"https://www.aibusiness.com/nlp/bud-e-aims-to-create-human-like-computer-voices"},{"dt":"February 13 2024","url":"https://www.heise.de/news/Bud-E-Laion-arbeitet-an-Audio-KI-mit-Empathie-9626286.html"},{"dt":"November 30 2023","url":"https://www.forbes.com/sites/hessiejones/2023/11/30/the-rise-and-fall-and-rise-of-sam-altman-has-grave-implications-for-ai-research-and-humanity/"},{"dt":"November 24 2023","url":"https://www.spektrum.de/news/interview-mit-laion-ki-im-sinn-der-menschheit-erforschen/2198369"},{"dt":"October 27 2023","url":"https://www.techcrunch.com/2023/10/27/a-group-behind-stable-diffusion-wants-to-open-source-emotion-detecting-ai"},{"dt":"September 29 2023","url":"https://www.golem.de/news/laion-riesiges-sprachmodell-fuer-deutsch-trainiert-2309-178081.html"},{"dt":"August 31 2023","url":"https://www.apnews.com/article/artists-ai-image-generators-stable-diffusion-midjourney-7ebcb6e6ddca3f165a3065c70ce85904"},{"dt":"July 16 2023","url":"https://www.ardmediathek.de/video/ttt-titel-thesen-temperamente/verliert-europa-den-anschluss/das-erste/Y3JpZDovL2Rhc2Vyc3RlLmRlL3R0dCAtIHRpdGVsIHRoZXNlbiB0ZW1wZXJhbWVudGUvYjMxZTE1NDAtM2M4OC00MGExLWJiNjQtODhlYTgzNzg0MzI5"},{"dt":"July 16 2023","url":"https://www.repubblica.it/economia/2023/07/17/news/maghi_tedeschi_algoritmi_ai-407769374/"},{"dt":"July 7 2023","url":"https://www.tagesschau.de/wissen/technologie/ki-trainingsdaten-privat-datenschutz-100.html"},{"dt":"June 15 2023","url":"https://www.golem.de/news/interview-mit-laion-der-datensatz-hinter-den-ki-bildern-2306-174786.html"},{"dt":"May 4 2023","url":"https://www.theguardian.com/technology/2023/may/04/eu-urged-to-protect-grassroots-ai-research-or-risk-losing-out-to-us"},{"dt":"April 28 2023","url":"https://www.unite.ai/laion-and-a-group-of-27/"},{"dt":"April 26 2023","url":"https://www.sifted.eu/articles/ai-supercomputer-petition-stable-diffusion"},{"dt":"April 24 2023","url":"https://www.bloomberg.com/news/features/2023-04-24/a-high-school-teacher-s-free-image-database-powers-ai-unicorns#xj4y7vzkg"},{"dt":"April 17 2023","url":"https://www.sueddeutsche.de/wirtschaft/ki-laion-kuenstliche-intelligenz-supercomputer-open-ai-chatgpt-1.5804840?reduced=true"},{"dt":"April 8 2023","url":"https://www.heise.de/news/Open-Source-KI-LAION-ruft-zu-offenem-GPT-4-Nachbau-auf-genug-Wissen-sei-da-8651446.html"},{"dt":"April 19 2023","url":"https://www.forbes.com/sites/hessiejones/2023/04/19/amid-growing-call-to-pause-ai-research-laion-petitions-governments-to-keep-agi-research-open-active-and-responsible/"},{"dt":"March 4 2023","url":"https://www.spiegel.de/spiegel/print/index-2023-10.html"},{"dt":"December 13 2022","url":"https://www.golem.de/news/laion-5b-datensatz-zum-ki-training-bietet-fast-6-milliarden-bilder-2212-170470.html"}]'),s=n(1280);function o(){return(0,r.jsxs)("div",{className:"w-full flex justify-center py-5 pt-16 md:pt-5",children:[(0,r.jsx)(i.Z,{title:"Press"}),(0,r.jsxs)("div",{className:"container px-5",children:[(0,r.jsx)("h1",{className:"text-7xl md:text-8xl font-bold pb-2",children:"PRESS RELEASES"}),(0,r.jsx)("hr",{className:"mb-5 mt-2 md:hidden"}),a.sort((function(e,t){return Date.parse((0,s.ZP)(t.dt))-Date.parse((0,s.ZP)(e.dt))})).map((function(e,t){return(0,r.jsxs)("div",{className:"bg-sky border border-paper mb-5 lg:flex-row p-5 shadow-lg shadow-neutral-800/20",children:[(0,r.jsxs)("p",{className:"text-3xl pb-2",children:[(0,s.ZP)(e.dt,"d mmmm, yyyy")," - ",(0,r.jsx)("b",{className:"tracking-wider",children:e.url.split("www.")[1].split("/")[0].toUpperCase()})]}),(0,r.jsx)("a",{className:"pt-3 break-all",href:e.url,children:e.url})]},t)}))]})]})}},9008:function(e,t,n){e.exports=n(3121)},1163:function(e,t,n){e.exports=n(880)},1280:function(e,t,n){"use strict";n.d(t,{ZP:function(){return s}});var r=/d{1,4}|D{3,4}|m{1,4}|yy(?:yy)?|([HhMsTt])\1?|W{1,2}|[LlopSZN]|"[^"]*"|'[^']*'/g,i=/\b(?:[A-Z]{1,3}[A-Z][TC])(?:[-+]\d{4})?|((?:Australian )?(?:Pacific|Mountain|Central|Eastern|Atlantic) (?:Standard|Daylight|Prevailing) Time)\b/g,a=/[^-+\dA-Z]/g;function s(e,t,n,i){if(1!==arguments.length||"string"!==typeof e||/\d/.test(e)||(t=e,e=void 0),(e=e||0===e?e:new Date)instanceof Date||(e=new Date(e)),isNaN(e))throw TypeError("Invalid date");var a=(t=String(o[t]||t||o.default)).slice(0,4);"UTC:"!==a&&"GMT:"!==a||(t=t.slice(4),n=!0,"GMT:"===a&&(i=!0));var s=function(){return n?"getUTC":"get"},h=function(){return e[s()+"Date"]()},p=function(){return e[s()+"Day"]()},w=function(){return e[s()+"Month"]()},g=function(){return e[s()+"FullYear"]()},y=function(){return e[s()+"Hours"]()},b=function(){return e[s()+"Minutes"]()},D=function(){return e[s()+"Seconds"]()},M=function(){return e[s()+"Milliseconds"]()},N=function(){return n?0:e.getTimezoneOffset()},x=function(){return d(e)},v=function(){return m(e)},T={d:function(){return h()},dd:function(){return l(h())},ddd:function(){return u.dayNames[p()]},DDD:function(){return c({y:g(),m:w(),d:h(),_:s(),dayName:u.dayNames[p()],short:!0})},dddd:function(){return u.dayNames[p()+7]},DDDD:function(){return c({y:g(),m:w(),d:h(),_:s(),dayName:u.dayNames[p()+7]})},m:function(){return w()+1},mm:function(){return l(w()+1)},mmm:function(){return u.monthNames[w()]},mmmm:function(){return u.monthNames[w()+12]},yy:function(){return String(g()).slice(2)},yyyy:function(){return l(g(),4)},h:function(){return y()%12||12},hh:function(){return l(y()%12||12)},H:function(){return y()},HH:function(){return l(y())},M:function(){return b()},MM:function(){return l(b())},s:function(){return D()},ss:function(){return l(D())},l:function(){return l(M(),3)},L:function(){return l(Math.floor(M()/10))},t:function(){return y()<12?u.timeNames[0]:u.timeNames[1]},tt:function(){return y()<12?u.timeNames[2]:u.timeNames[3]},T:function(){return y()<12?u.timeNames[4]:u.timeNames[5]},TT:function(){return y()<12?u.timeNames[6]:u.timeNames[7]},Z:function(){return i?"GMT":n?"UTC":f(e)},o:function(){return(N()>0?"-":"+")+l(100*Math.floor(Math.abs(N())/60)+Math.abs(N())%60,4)},p:function(){return(N()>0?"-":"+")+l(Math.floor(Math.abs(N())/60),2)+":"+l(Math.floor(Math.abs(N())%60),2)},S:function(){return["th","st","nd","rd"][h()%10>3?0:(h()%100-h()%10!=10)*h()%10]},W:function(){return x()},WW:function(){return l(x())},N:function(){return v()}};return t.replace(r,(function(e){return e in T?T[e]():e.slice(1,e.length-1)}))}var o={default:"ddd mmm dd yyyy HH:MM:ss",shortDate:"m/d/yy",paddedShortDate:"mm/dd/yyyy",mediumDate:"mmm d, yyyy",longDate:"mmmm d, yyyy",fullDate:"dddd, mmmm d, yyyy",shortTime:"h:MM TT",mediumTime:"h:MM:ss TT",longTime:"h:MM:ss TT Z",isoDate:"yyyy-mm-dd",isoTime:"HH:MM:ss",isoDateTime:"yyyy-mm-dd'T'HH:MM:sso",isoUtcDateTime:"UTC:yyyy-mm-dd'T'HH:MM:ss'Z'",expiresHeaderFormat:"ddd, dd mmm yyyy HH:MM:ss Z"},u={dayNames:["Sun","Mon","Tue","Wed","Thu","Fri","Sat","Sunday","Monday","Tuesday","Wednesday","Thursday","Friday","Saturday"],monthNames:["Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep","Oct","Nov","Dec","January","February","March","April","May","June","July","August","September","October","November","December"],timeNames:["a","p","am","pm","A","P","AM","PM"]},l=function(e){var t=arguments.length>1&&void 0!==arguments[1]?arguments[1]:2;return String(e).padStart(t,"0")},c=function(e){var t=e.y,n=e.m,r=e.d,i=e._,a=e.dayName,s=e.short,o=void 0!==s&&s,u=new Date,l=new Date;l.setDate(l[i+"Date"]()-1);var c=new Date;c.setDate(c[i+"Date"]()+1);return u[i+"FullYear"]()===t&&u[i+"Month"]()===n&&u[i+"Date"]()===r?o?"Tdy":"Today":l[i+"FullYear"]()===t&&l[i+"Month"]()===n&&l[i+"Date"]()===r?o?"Ysd":"Yesterday":c[i+"FullYear"]()===t&&c[i+"Month"]()===n&&c[i+"Date"]()===r?o?"Tmw":"Tomorrow":a},d=function(e){var t=new Date(e.getFullYear(),e.getMonth(),e.getDate());t.setDate(t.getDate()-(t.getDay()+6)%7+3);var n=new Date(t.getFullYear(),0,4);n.setDate(n.getDate()-(n.getDay()+6)%7+3);var r=t.getTimezoneOffset()-n.getTimezoneOffset();t.setHours(t.getHours()-r);var i=(t-n)/6048e5;return 1+Math.floor(i)},m=function(e){var t=e.getDay();return 0===t&&(t=7),t},f=function(e){return(String(e).match(i)||[""]).pop().replace(a,"").replace(/GMT\+0000/g,"UTC")}}},function(e){e.O(0,[774,888,179],(function(){return t=8923,e(e.s=t);var t}));var t=e.O();_N_E=t}]); \ No newline at end of file diff --git a/_next/static/chunks/pages/press-f2724375625fc903.js b/_next/static/chunks/pages/press-f2724375625fc903.js deleted file mode 100644 index 59dcda03..00000000 --- a/_next/static/chunks/pages/press-f2724375625fc903.js +++ /dev/null @@ -1 +0,0 @@ -(self.webpackChunk_N_E=self.webpackChunk_N_E||[]).push([[313],{8923:function(e,t,n){(window.__NEXT_P=window.__NEXT_P||[]).push(["/press",function(){return n(817)}])},8396:function(e,t,n){"use strict";n.d(t,{Z:function(){return o}});var r=n(5893),i=n(9008),a=n.n(i),s=n(1163);function o(e){var t=(0,s.useRouter)(),n=e.title?e.title+" | LAION":"LAION",i=e.desc?e.desc:"LAION, Large-scale Artificial Intelligence Open Network, is a non-profit organization making machine learning resources available to the general public.",o=e.image?e.image:"/social.png",u=e.alt?e.alt:"The text: LAION. Large-scale Artificial Intelligence Open Network, TRULY OPEN AI. 100% NON-PROFIT. 100% FREE.",l=e.slug?e.slug:t.route;return(0,r.jsxs)(a(),{children:[(0,r.jsx)("title",{children:n}),(0,r.jsx)("meta",{name:"title",content:n}),(0,r.jsx)("meta",{property:"og:title",content:n}),(0,r.jsx)("meta",{name:"twitter:title",content:n}),(0,r.jsx)("meta",{name:"description",content:i}),(0,r.jsx)("meta",{property:"og:description",content:i}),(0,r.jsx)("meta",{name:"twitter:description",content:i}),(0,r.jsx)("meta",{property:"og:image",content:"https://laion.ai"+o}),(0,r.jsx)("meta",{name:"twitter:image",content:"https://laion.ai"+o}),(0,r.jsx)("meta",{name:"twitter:image:alt",content:u}),(0,r.jsx)("meta",{property:"og:type",content:"website"}),(0,r.jsx)("meta",{property:"og:url",content:"https://laion.ai"+l}),(0,r.jsx)("meta",{name:"twitter:url",content:"https://laion.ai"+l}),(0,r.jsx)("meta",{name:"twitter:card",content:"summary_large_image"}),(0,r.jsx)("meta",{name:"viewport",content:"initial-scale=1.0, width=device-width"}),(0,r.jsx)("meta",{name:"theme-color",content:"#1D374E"}),(0,r.jsx)("link",{rel:"icon",type:"image/png",sizes:"32x32",href:"/favicon.png"}),(0,r.jsx)("link",{rel:"icon",href:"/favicon.svg",type:"image/svg+xml"}),(0,r.jsx)("link",{rel:"preload",href:"/fonts/DinishCondensed-Bold.woff",as:"font",type:"font/woff",crossOrigin:"true"}),(0,r.jsx)("link",{rel:"preload",href:"/fonts/DinishCondensed-Bold.woff2",as:"font",type:"font/woff2",crossOrigin:"true"}),(0,r.jsx)("link",{rel:"preload",href:"/fonts/Dinish-Regular.woff",as:"font",type:"font/woff",crossOrigin:"true"}),(0,r.jsx)("link",{rel:"preload",href:"/fonts/Dinish-Regular.woff2",as:"font",type:"font/woff2",crossOrigin:"true"}),(0,r.jsx)("link",{rel:"preload",href:"/fonts/Dinish-Italic.woff",as:"font",type:"font/woff",crossOrigin:"true"}),(0,r.jsx)("link",{rel:"preload",href:"/fonts/Dinish-Italic.woff2",as:"font",type:"font/woff2",crossOrigin:"true"})]})}},817:function(e,t,n){"use strict";n.r(t),n.d(t,{default:function(){return o}});var r=n(5893),i=n(8396),a=JSON.parse('[{"dt":"August 30 2024","url":"https://www.arstechnica.com/tech-policy/2024/08/nonprofit-scrubs-illegal-content-from-controversial-ai-training-dataset"},{"dt":"June 10 2024","url":"https://www.heise.de/news/Reasoning-Fail-Gaengige-LLMs-scheitern-an-kinderleichter-Aufgabe-9755034.html"},{"dt":"February 16 2024","url":"https://www.nature.com/articles/d41586-024-00497-8"},{"dt":"February 15 2024","url":"https://www.techcrunch.com/2024/02/15/this-german-nonprofit-is-building-an-open-voice-assistant-that-anyone-can-use/"},{"dt":"February 14 2024","url":"https://www.heise.de/news/KI-Update-kompakt-ChatGPT-personalisieren-Digitalisierung-IAFM-Bud-E-9628384.html"},{"dt":"February 13 2024","url":"https://www.aibusiness.com/nlp/bud-e-aims-to-create-human-like-computer-voices"},{"dt":"February 13 2024","url":"https://www.heise.de/news/Bud-E-Laion-arbeitet-an-Audio-KI-mit-Empathie-9626286.html"},{"dt":"November 30 2023","url":"https://www.forbes.com/sites/hessiejones/2023/11/30/the-rise-and-fall-and-rise-of-sam-altman-has-grave-implications-for-ai-research-and-humanity/"},{"dt":"November 24 2023","url":"https://www.spektrum.de/news/interview-mit-laion-ki-im-sinn-der-menschheit-erforschen/2198369"},{"dt":"October 27 2023","url":"https://www.techcrunch.com/2023/10/27/a-group-behind-stable-diffusion-wants-to-open-source-emotion-detecting-ai"},{"dt":"September 29 2023","url":"https://www.golem.de/news/laion-riesiges-sprachmodell-fuer-deutsch-trainiert-2309-178081.html"},{"dt":"August 31 2023","url":"https://www.apnews.com/article/artists-ai-image-generators-stable-diffusion-midjourney-7ebcb6e6ddca3f165a3065c70ce85904"},{"dt":"July 16 2023","url":"https://www.ardmediathek.de/video/ttt-titel-thesen-temperamente/verliert-europa-den-anschluss/das-erste/Y3JpZDovL2Rhc2Vyc3RlLmRlL3R0dCAtIHRpdGVsIHRoZXNlbiB0ZW1wZXJhbWVudGUvYjMxZTE1NDAtM2M4OC00MGExLWJiNjQtODhlYTgzNzg0MzI5"},{"dt":"July 16 2023","url":"https://www.repubblica.it/economia/2023/07/17/news/maghi_tedeschi_algoritmi_ai-407769374/"},{"dt":"July 7 2023","url":"https://www.tagesschau.de/wissen/technologie/ki-trainingsdaten-privat-datenschutz-100.html"},{"dt":"June 15 2023","url":"https://www.golem.de/news/interview-mit-laion-der-datensatz-hinter-den-ki-bildern-2306-174786.html"},{"dt":"May 4 2023","url":"https://www.theguardian.com/technology/2023/may/04/eu-urged-to-protect-grassroots-ai-research-or-risk-losing-out-to-us"},{"dt":"April 28 2023","url":"https://www.unite.ai/laion-and-a-group-of-27/"},{"dt":"April 26 2023","url":"https://www.sifted.eu/articles/ai-supercomputer-petition-stable-diffusion"},{"dt":"April 24 2023","url":"https://www.bloomberg.com/news/features/2023-04-24/a-high-school-teacher-s-free-image-database-powers-ai-unicorns#xj4y7vzkg"},{"dt":"April 17 2023","url":"https://www.sueddeutsche.de/wirtschaft/ki-laion-kuenstliche-intelligenz-supercomputer-open-ai-chatgpt-1.5804840?reduced=true"},{"dt":"April 8 2023","url":"https://www.heise.de/news/Open-Source-KI-LAION-ruft-zu-offenem-GPT-4-Nachbau-auf-genug-Wissen-sei-da-8651446.html"},{"dt":"April 19 2023","url":"https://www.forbes.com/sites/hessiejones/2023/04/19/amid-growing-call-to-pause-ai-research-laion-petitions-governments-to-keep-agi-research-open-active-and-responsible/"},{"dt":"March 4 2023","url":"https://www.spiegel.de/spiegel/print/index-2023-10.html"},{"dt":"December 13 2022","url":"https://www.golem.de/news/laion-5b-datensatz-zum-ki-training-bietet-fast-6-milliarden-bilder-2212-170470.html"}]'),s=n(1280);function o(){return(0,r.jsxs)("div",{className:"w-full flex justify-center py-5 pt-16 md:pt-5",children:[(0,r.jsx)(i.Z,{title:"Press"}),(0,r.jsxs)("div",{className:"container px-5",children:[(0,r.jsx)("h1",{className:"text-7xl md:text-8xl font-bold pb-2",children:"PRESS RELEASES"}),(0,r.jsx)("hr",{className:"mb-5 mt-2 md:hidden"}),a.sort((function(e,t){return Date.parse((0,s.ZP)(t.dt))-Date.parse((0,s.ZP)(e.dt))})).map((function(e,t){return(0,r.jsxs)("div",{className:"bg-sky border border-paper mb-5 lg:flex-row p-5 shadow-lg shadow-neutral-800/20",children:[(0,r.jsxs)("p",{className:"text-3xl pb-2",children:[(0,s.ZP)(e.dt,"d mmmm, yyyy")," - ",(0,r.jsx)("b",{className:"tracking-wider",children:e.url.split("www.")[1].split("/")[0].toUpperCase()})]}),(0,r.jsx)("a",{className:"pt-3 break-all",href:e.url,children:e.url})]},t)}))]})]})}},9008:function(e,t,n){e.exports=n(3121)},1163:function(e,t,n){e.exports=n(880)},1280:function(e,t,n){"use strict";n.d(t,{ZP:function(){return s}});var r=/d{1,4}|D{3,4}|m{1,4}|yy(?:yy)?|([HhMsTt])\1?|W{1,2}|[LlopSZN]|"[^"]*"|'[^']*'/g,i=/\b(?:[A-Z]{1,3}[A-Z][TC])(?:[-+]\d{4})?|((?:Australian )?(?:Pacific|Mountain|Central|Eastern|Atlantic) (?:Standard|Daylight|Prevailing) Time)\b/g,a=/[^-+\dA-Z]/g;function s(e,t,n,i){if(1!==arguments.length||"string"!==typeof e||/\d/.test(e)||(t=e,e=void 0),(e=e||0===e?e:new Date)instanceof Date||(e=new Date(e)),isNaN(e))throw TypeError("Invalid date");var a=(t=String(o[t]||t||o.default)).slice(0,4);"UTC:"!==a&&"GMT:"!==a||(t=t.slice(4),n=!0,"GMT:"===a&&(i=!0));var s=function(){return n?"getUTC":"get"},h=function(){return e[s()+"Date"]()},p=function(){return e[s()+"Day"]()},w=function(){return e[s()+"Month"]()},g=function(){return e[s()+"FullYear"]()},y=function(){return e[s()+"Hours"]()},b=function(){return e[s()+"Minutes"]()},D=function(){return e[s()+"Seconds"]()},M=function(){return e[s()+"Milliseconds"]()},N=function(){return n?0:e.getTimezoneOffset()},x=function(){return d(e)},v=function(){return m(e)},T={d:function(){return h()},dd:function(){return l(h())},ddd:function(){return u.dayNames[p()]},DDD:function(){return c({y:g(),m:w(),d:h(),_:s(),dayName:u.dayNames[p()],short:!0})},dddd:function(){return u.dayNames[p()+7]},DDDD:function(){return c({y:g(),m:w(),d:h(),_:s(),dayName:u.dayNames[p()+7]})},m:function(){return w()+1},mm:function(){return l(w()+1)},mmm:function(){return u.monthNames[w()]},mmmm:function(){return u.monthNames[w()+12]},yy:function(){return String(g()).slice(2)},yyyy:function(){return l(g(),4)},h:function(){return y()%12||12},hh:function(){return l(y()%12||12)},H:function(){return y()},HH:function(){return l(y())},M:function(){return b()},MM:function(){return l(b())},s:function(){return D()},ss:function(){return l(D())},l:function(){return l(M(),3)},L:function(){return l(Math.floor(M()/10))},t:function(){return y()<12?u.timeNames[0]:u.timeNames[1]},tt:function(){return y()<12?u.timeNames[2]:u.timeNames[3]},T:function(){return y()<12?u.timeNames[4]:u.timeNames[5]},TT:function(){return y()<12?u.timeNames[6]:u.timeNames[7]},Z:function(){return i?"GMT":n?"UTC":f(e)},o:function(){return(N()>0?"-":"+")+l(100*Math.floor(Math.abs(N())/60)+Math.abs(N())%60,4)},p:function(){return(N()>0?"-":"+")+l(Math.floor(Math.abs(N())/60),2)+":"+l(Math.floor(Math.abs(N())%60),2)},S:function(){return["th","st","nd","rd"][h()%10>3?0:(h()%100-h()%10!=10)*h()%10]},W:function(){return x()},WW:function(){return l(x())},N:function(){return v()}};return t.replace(r,(function(e){return e in T?T[e]():e.slice(1,e.length-1)}))}var o={default:"ddd mmm dd yyyy HH:MM:ss",shortDate:"m/d/yy",paddedShortDate:"mm/dd/yyyy",mediumDate:"mmm d, yyyy",longDate:"mmmm d, yyyy",fullDate:"dddd, mmmm d, yyyy",shortTime:"h:MM TT",mediumTime:"h:MM:ss TT",longTime:"h:MM:ss TT Z",isoDate:"yyyy-mm-dd",isoTime:"HH:MM:ss",isoDateTime:"yyyy-mm-dd'T'HH:MM:sso",isoUtcDateTime:"UTC:yyyy-mm-dd'T'HH:MM:ss'Z'",expiresHeaderFormat:"ddd, dd mmm yyyy HH:MM:ss Z"},u={dayNames:["Sun","Mon","Tue","Wed","Thu","Fri","Sat","Sunday","Monday","Tuesday","Wednesday","Thursday","Friday","Saturday"],monthNames:["Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep","Oct","Nov","Dec","January","February","March","April","May","June","July","August","September","October","November","December"],timeNames:["a","p","am","pm","A","P","AM","PM"]},l=function(e){var t=arguments.length>1&&void 0!==arguments[1]?arguments[1]:2;return String(e).padStart(t,"0")},c=function(e){var t=e.y,n=e.m,r=e.d,i=e._,a=e.dayName,s=e.short,o=void 0!==s&&s,u=new Date,l=new Date;l.setDate(l[i+"Date"]()-1);var c=new Date;c.setDate(c[i+"Date"]()+1);return u[i+"FullYear"]()===t&&u[i+"Month"]()===n&&u[i+"Date"]()===r?o?"Tdy":"Today":l[i+"FullYear"]()===t&&l[i+"Month"]()===n&&l[i+"Date"]()===r?o?"Ysd":"Yesterday":c[i+"FullYear"]()===t&&c[i+"Month"]()===n&&c[i+"Date"]()===r?o?"Tmw":"Tomorrow":a},d=function(e){var t=new Date(e.getFullYear(),e.getMonth(),e.getDate());t.setDate(t.getDate()-(t.getDay()+6)%7+3);var n=new Date(t.getFullYear(),0,4);n.setDate(n.getDate()-(n.getDay()+6)%7+3);var r=t.getTimezoneOffset()-n.getTimezoneOffset();t.setHours(t.getHours()-r);var i=(t-n)/6048e5;return 1+Math.floor(i)},m=function(e){var t=e.getDay();return 0===t&&(t=7),t},f=function(e){return(String(e).match(i)||[""]).pop().replace(a,"").replace(/GMT\+0000/g,"UTC")}}},function(e){e.O(0,[774,888,179],(function(){return t=8923,e(e.s=t);var t}));var t=e.O();_N_E=t}]); \ No newline at end of file diff --git a/about/index.html b/about/index.html index 7f6a2228..7ae3472e 100644 --- a/about/index.html +++ b/about/index.html @@ -1,8 +1,8 @@ -About | LAION

LAION

ABOUT

We 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.

+About | LAION

ABOUT

We 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.

-
\ No newline at end of file +
\ No newline at end of file diff --git a/blog-de/index.html b/blog-de/index.html index 79bd61f8..a9ca0958 100644 --- a/blog-de/index.html +++ b/blog-de/index.html @@ -1,2 +1,2 @@ -Blog | LAION

LAION

BLOG

Willkommen in unserem LAION-Blog! Hier finden Sie Kommentare, Nachrichten und Updates zu unseren aktuellen Forschungsprojekten und Fortschritten im Bereich der KI-Forschung. Diese Blogbeiträge sind nicht als vollständige wissenschaftliche Forschungsarbeiten gedacht, sondern als Arbeitsfortschritte, um weitere Forschungen/Diskussionen auf unserem Discord-Server und in der offenen wissenschaftlichen Gemeinschaft zu fördern.

LeoLM: Ein Impuls für Deutschsprachige LLM-Forschung

by: Björn Plüster, 28 Sep, 2023

Lernen Sie LeoLM kennen, das erste offen und kommerziell verfügbare deutsche Foundation Language Model, das auf Llama-2 basiert. -Unsere Modelle erweitern die Fähigkeiten von Llama-2 durch ein fortgesetztes Training auf einem großen Korpus von hochwertigen deutschen und größtenteils lokal spezifische...

\ No newline at end of file +Blog | LAION

LAION

BLOG

Willkommen in unserem LAION-Blog! Hier finden Sie Kommentare, Nachrichten und Updates zu unseren aktuellen Forschungsprojekten und Fortschritten im Bereich der KI-Forschung. Diese Blogbeiträge sind nicht als vollständige wissenschaftliche Forschungsarbeiten gedacht, sondern als Arbeitsfortschritte, um weitere Forschungen/Diskussionen auf unserem Discord-Server und in der offenen wissenschaftlichen Gemeinschaft zu fördern.

LeoLM: Ein Impuls für Deutschsprachige LLM-Forschung

by: Björn Plüster, 28 Sep, 2023

Lernen Sie LeoLM kennen, das erste offen und kommerziell verfügbare deutsche Foundation Language Model, das auf Llama-2 basiert. +Unsere Modelle erweitern die Fähigkeiten von Llama-2 durch ein fortgesetztes Training auf einem großen Korpus von hochwertigen deutschen und größtenteils lokal spezifische...

\ No newline at end of file diff --git a/blog-de/leo-lm/index.html b/blog-de/leo-lm/index.html index 0c40aab4..3abece48 100644 --- a/blog-de/leo-lm/index.html +++ b/blog-de/leo-lm/index.html @@ -1,7 +1,7 @@ LeoLM: Ein Impuls für Deutschsprachige LLM-Forschung | LAION

LAION

LEOLM: EIN IMPULS FÜR DEUTSCHSPRACHIGE LLM-FORSCHUNG

by: Björn Plüster, 28 Sep, 2023

Lernen Sie LeoLM kennen, das erste offen und kommerziell verfügbare deutsche Foundation Language Model, das auf Llama-2 basiert. +Unsere Modelle erweitern ..."/>

LEOLM: EIN IMPULS FÜR DEUTSCHSPRACHIGE LLM-FORSCHUNG

by: Björn Plüster, 28 Sep, 2023

Lernen Sie LeoLM kennen, das erste offen und kommerziell verfügbare deutsche Foundation Language Model, das auf Llama-2 basiert. Unsere Modelle erweitern die Fähigkeiten von Llama-2 durch ein fortgesetztes Training auf einem großen Korpus von hochwertigen deutschen und größtenteils lokal spezifischen Texten. Dank eines Compute-Grants auf dem neuen Supercomputer 42 von HessianAI veröffentlichen wir zwei Foundation-Modelle, die mit einer Kontextlänge von 8k trainiert wurden, LeoLM/leo-hessianai-7b und LeoLM/leo-hessianai-13b (70b folgt auch bald! 👀) unter der Llama-2 Community-Lizenz. Zusätzlich konstruieren wir einen Evaluierungssatz für Benchmarks zur Überprüfung der Fähigkeiten deutscher Sprachmodelle, um den Modellvergleich zu standardisieren, ähnlich zu den weit verbreiteten auf Englisch basierten Evaluierungen, wie sie beispielsweise von lm-evaluation-harness oder LLM-Foundry bereitgestellt werden. @@ -98,4 +98,4 @@

BUD-E: ENHANCING AI VOICE ASSISTANTS’ CONVERSATIONAL QUALITY, NATURALNESS AND EMPATHY

by: LAION, 08 Feb, 2024

AI voice assistants have revolutionized our interaction with technology, answering queries, performing tasks, and making life easier. However, the stilted, mechanical nature of their responses is a barrier to truly immersive conversational experiences. Unlike human conversation partners, they often struggle with fully understanding and adapting to the nuanced, emotional, and contextually rich nature of human dialogue, leading to noticeable latencies and a disjointed conversational flow. Consequently, users often experience unsatisfactory exchanges, lacking emotional resonance and context familiarity.

+BUD-E: Enhancing AI Voice Assistants’ Conversational Quality, Naturalness and Empathy | LAION

BUD-E: ENHANCING AI VOICE ASSISTANTS’ CONVERSATIONAL QUALITY, NATURALNESS AND EMPATHY

by: LAION, 08 Feb, 2024

AI voice assistants have revolutionized our interaction with technology, answering queries, performing tasks, and making life easier. However, the stilted, mechanical nature of their responses is a barrier to truly immersive conversational experiences. Unlike human conversation partners, they often struggle with fully understanding and adapting to the nuanced, emotional, and contextually rich nature of human dialogue, leading to noticeable latencies and a disjointed conversational flow. Consequently, users often experience unsatisfactory exchanges, lacking emotional resonance and context familiarity.

BUD-E

Wouldn’t it be awesome to have a fully open voice assistant that can

    @@ -62,4 +62,4 @@

    Collaborating to Build the Future of Conversational AI

    The development of BUD-E is an ongoing process that requires the collective effort of a diverse community. We invite open-source developers, researchers, and enthusiasts to join us in refining BUD-E's individual modules and contributing to its growth. Together, we can create an AI voice assistants that engage with us in natural, intuitive, and empathetic conversations.

    If you're interested in contributing to this project, join our Discord community or reach out to us at bud-e@laion.ai.

    -
\ No newline at end of file +
\ No newline at end of file diff --git a/blog/clara-release/index.html b/blog/clara-release/index.html index 772a5c00..ee22ecab 100644 --- a/blog/clara-release/index.html +++ b/blog/clara-release/index.html @@ -1,4 +1,4 @@ -CLARA: Advancing Machines in Understanding Speech Nuances | LAION

LAION

CLARA: ADVANCING MACHINES IN UNDERSTANDING SPEECH NUANCES

by: Knoriy, Christoph, Robert, 16 Oct, 2023

Voices carry not only words but also convey emotions, emphasis, and nuance through aspects like tone and accent. However, existing speech technology only partially comprehends these intricate components of human speech.

+CLARA: Advancing Machines in Understanding Speech Nuances | LAION

CLARA: ADVANCING MACHINES IN UNDERSTANDING SPEECH NUANCES

by: Knoriy, Christoph, Robert, 16 Oct, 2023

Voices carry not only words but also convey emotions, emphasis, and nuance through aspects like tone and accent. However, existing speech technology only partially comprehends these intricate components of human speech.

Introducing CLARA (Multilingual Contrastive Learning for Audio Representation Acquisition) – a project designed to enhance machines' understanding of the implicit aspects of speech. It aspires to develop a comprehensive pre-trained model dedicated to auditory communication.

@@ -32,4 +32,4 @@

Acknowledgement

We would like to thank Stability AI for their generous support in providing the essential compute resources for this project.

- \ No newline at end of file + \ No newline at end of file diff --git a/blog/coca/index.html b/blog/coca/index.html index 088404de..6201f06e 100644 --- a/blog/coca/index.html +++ b/blog/coca/index.html @@ -1,4 +1,4 @@ -Training Contrastive Captioners | LAION

TRAINING CONTRASTIVE CAPTIONERS

by: Giovanni Puccetti, Maciej Kilian, Romain Beaumont, 02 Feb, 2023

We introduce a new model type to OpenClip Contrastive Captioners (CoCa) [1]. This model adds an autoregressive objective (generation) on top of the CLIP contrastive one. The architecture is composed of three parts, the first two are similar to those composing a CLIP model and the third is a text decoder that stands on top of the text encoder. The additional decoder takes as input the encoded images (through cross-attention) and the previous tokens to predict the next most probable one. One of the few architecture changes, compared to CLIP, is attentional pooling [2], used to aggregate image representations and pass them to both the contrastive loss and the decoder cross-attention.

+Training Contrastive Captioners | LAION

TRAINING CONTRASTIVE CAPTIONERS

by: Giovanni Puccetti, Maciej Kilian, Romain Beaumont, 02 Feb, 2023

We introduce a new model type to OpenClip Contrastive Captioners (CoCa) [1]. This model adds an autoregressive objective (generation) on top of the CLIP contrastive one. The architecture is composed of three parts, the first two are similar to those composing a CLIP model and the third is a text decoder that stands on top of the text encoder. The additional decoder takes as input the encoded images (through cross-attention) and the previous tokens to predict the next most probable one. One of the few architecture changes, compared to CLIP, is attentional pooling [2], used to aggregate image representations and pass them to both the contrastive loss and the decoder cross-attention.

This is interesting for several reasons:

ANNOUNCING DATACOMP: IN SEARCH OF THE NEXT GENERATION OF MULTIMODAL DATASETS

by: Gabriel Ilharco, 27 Apr, 2023

[ Paper ] [ Code ] [ Website ]

+Announcing DataComp: In search of the next generation of multimodal datasets | LAION

ANNOUNCING DATACOMP: IN SEARCH OF THE NEXT GENERATION OF MULTIMODAL DATASETS

by: Gabriel Ilharco, 27 Apr, 2023

[ Paper ] [ Code ] [ Website ]

About a year ago, we released LAION-5B, a billion-scale open-source image-text dataset. Since then, LAION-5B has become a staple in the open-source machine learning ecosystem, powering open-source models like OpenCLIP, OpenFlamingo, and Stable Diffusion. From the beginning, we viewed LAION-5B as only the first step on this research journey and hoped that we can build the next generation of multimodal datasets both rigorously and collaboratively in the open as a research community.

Today, we are proud to introduce DataComp, a new benchmark for designing multimodal datasets. Unlike traditional benchmarks focused on modeling improvements, DataComp puts data front and center. In Datacomp, participants innovate by proposing new training sets, leaving the training code, hyper-parameters and compute fixed. As part of our competition, we are releasing CommonPool, the largest public collection of image-text pairs to date with 12.8B samples.

Along with our pool, we also release DataComp-1B, a 1.4B subset that can be used to outperform compute-matched CLIP models from OpenAI and LAION. DataComp-1B makes it possible to train a CLIP ViT-L model to better performance than a larger ViT-g model trained on LAION-2B while using 9x less training compute. Our ViT-L/14 trained on DataComp-1B obtains 79.2% zero-shot accuracy on ImageNet, substantially outperforming OpenAI's model trained with the same compute (75.5% zero-shot accuracy).

@@ -81,4 +81,4 @@

We thank all of our paper authors: Samir Gadre, Gabriel Ilharco, Alex Fang, Jonathan Hayase, Georgios Smyrnis, Thao Nguyen, Ryan Marten, Mitchell Wortsman, Dhruba Ghosh, Jieyu Zhang, Eyal Orgad, Rahim Entezari, Giannis Daras, Sarah Pratt, Vivek Ramanujan, Yonatan Bitton, Kalyani Marathe, Stephen Mussmann, Richard Vencu, Mehdi Cherti, Ranjay Krishna, Pang Wei Koh, Olga Saukh, Alexander Ratner, Shuran Song, Hannaneh Hajishirzi, Ali Farhadi, Romain Beaumont, Sewoong Oh, Alex Dimakis, Jenia Jitsev, Yair Carmon, Vaishaal Shankar, Ludwig Schmidt.

We also thank Amro Abbas, Jessie Chapman, Brian Cheung, Joshua Gardner, Nancy Garland, Sachin Goyal, Huy Ha, Zaid Harchaoui, Andy Jones, Adam Klivans, Daniel Levy, Ronak Mehta, Ari Morcos, Raviteja Mullapudi, Kentrell Owens, Alec Radford, Marco Tulio Ribeiro, Shiori Sagawa, Christoph Schuhmann, Matthew Wallingford, and Ross Wightman for helpful feedback at various stages of the project.

A special thanks to Stability AI and the Gauss Centre for Supercomputing e.V (compute time granted on JUWELS Booster hosted at Juelich Supercomputing Center) for providing us with compute resources to train models, without which none of this would have been possible.

-

\ No newline at end of file +
\ No newline at end of file diff --git a/blog/falling-walls-2023/index.html b/blog/falling-walls-2023/index.html index ac354e22..4ded4c30 100644 --- a/blog/falling-walls-2023/index.html +++ b/blog/falling-walls-2023/index.html @@ -1,4 +1,4 @@ -LAION Triumphs at the Falling Walls Science Breakthrough of the Year 2023 Awards | LAION

LAION TRIUMPHS AT THE FALLING WALLS SCIENCE BREAKTHROUGH OF THE YEAR 2023 AWARDS

by: Christoph, Jenia, Robert, 14 Sep, 2023

We happily announce that we, LAION, won the Falling Walls Science Breakthrough of the Year 2023 Award in the category Science and Innovation Management for "democratizing AI research by providing open access to advanced AI models, tools, and datasets, fostering public engagement and awareness, and promoting international collaboration to create a transparent and inclusive AI ecosystem that benefits everyone." This recognition is not just for the select few but for our entire LAION community of hobby scientists, university professors, students, and enthusiasts all united with a shared vision - the democratization of AI research.

+LAION Triumphs at the Falling Walls Science Breakthrough of the Year 2023 Awards | LAION

LAION TRIUMPHS AT THE FALLING WALLS SCIENCE BREAKTHROUGH OF THE YEAR 2023 AWARDS

by: Christoph, Jenia, Robert, 14 Sep, 2023

We happily announce that we, LAION, won the Falling Walls Science Breakthrough of the Year 2023 Award in the category Science and Innovation Management for "democratizing AI research by providing open access to advanced AI models, tools, and datasets, fostering public engagement and awareness, and promoting international collaboration to create a transparent and inclusive AI ecosystem that benefits everyone." This recognition is not just for the select few but for our entire LAION community of hobby scientists, university professors, students, and enthusiasts all united with a shared vision - the democratization of AI research.

FW23A

About the Award

The Falling Walls Science Breakthrough of the Year Award, bestowed by the non-profit Falling Walls Foundation, celebrates groundbreaking achievements across various academic disciplines. Inspired by the historic fall of the Berlin Wall, it embodies the spirit of overcoming barriers to forge a brighter future for society.

@@ -10,4 +10,4 @@

We extend our heartfelt gratitude to the Falling Walls Foundation for acknowledging our efforts and to every member of the LAION community for being an indispensable part of this journey.

As we celebrate this monumental achievement, we invite you to become a part of this vibrant community. Together, we can foster a culture of open dialogue, collaboration, and innovative solutions. Join us on Discord as we continue to break walls and democratize AI research for the betterment of society globally.

For more details on the award, visit the official announcement.

-

\ No newline at end of file +
\ No newline at end of file diff --git a/blog/giant-openclip/index.html b/blog/giant-openclip/index.html index 2c89b7d1..da83b8e5 100644 --- a/blog/giant-openclip/index.html +++ b/blog/giant-openclip/index.html @@ -1,4 +1,4 @@ -Reaching 80% zero-shot accuracy with OpenCLIP: ViT-G/14 trained on LAION-2B | LAION

REACHING 80% ZERO-SHOT ACCURACY WITH OPENCLIP: VIT-G/14 TRAINED ON LAION-2B

by: Mitchell Wortsman, 24 Jan, 2023

We have trained a new ViT-G/14 CLIP model with OpenCLIP which achieves 80.1% zero-shot accuracy on ImageNet and 74.9% zero-shot image retrieval (Recall@5) on MS COCO. As of January 2023, this is the best open source CLIP model.

+Reaching 80% zero-shot accuracy with OpenCLIP: ViT-G/14 trained on LAION-2B | LAION

REACHING 80% ZERO-SHOT ACCURACY WITH OPENCLIP: VIT-G/14 TRAINED ON LAION-2B

by: Mitchell Wortsman, 24 Jan, 2023

We have trained a new ViT-G/14 CLIP model with OpenCLIP which achieves 80.1% zero-shot accuracy on ImageNet and 74.9% zero-shot image retrieval (Recall@5) on MS COCO. As of January 2023, this is the best open source CLIP model.

We believe this is interesting because:

  • CLIP models are useful for zero-shot classification, retrieval, and for guidance/conditioning in generative models (OpenCLIP is used in Stable Diffusion V2 and currently the third most downloaded model on HuggingFace is a CLIP model). The approach underlying CLIP—self supervised learning on a large, heterogeneous dataset—has been shown to produce models which are more robust and fair.
    • @@ -186,4 +186,4 @@

    Ludwig Schmidt and Ali Farhadi for helpful discussions, and to the RAIVN and EFML labs at the University of Washington

And of course thanks to Emad and Stability AI for providing the compute resources used for these experiments.

-
\ No newline at end of file +
\ No newline at end of file diff --git a/blog/h14_clip_retrieval/index.html b/blog/h14_clip_retrieval/index.html index a33fb12a..74750f0e 100644 --- a/blog/h14_clip_retrieval/index.html +++ b/blog/h14_clip_retrieval/index.html @@ -1,4 +1,4 @@ -Clip-Retrieval Update: H-14 Index & SLURM Inference | LAION

CLIP-RETRIEVAL UPDATE: H-14 INDEX & SLURM INFERENCE

by: no usr, 31 Jan, 2023

Today we release a KNN index for LAION-5B that allows for fast queries of the dataset with the open clip ViT-H-14 CLIP model. This means that users can search through billions of samples quickly and easily, making it a powerful tool for various applications such as image and text retrieval, data filtering and more. With this update also comes a brand new SLURM based inference backend for high-compute environments.

+Clip-Retrieval Update: H-14 Index & SLURM Inference | LAION

CLIP-RETRIEVAL UPDATE: H-14 INDEX & SLURM INFERENCE

by: no usr, 31 Jan, 2023

Today we release a KNN index for LAION-5B that allows for fast queries of the dataset with the open clip ViT-H-14 CLIP model. This means that users can search through billions of samples quickly and easily, making it a powerful tool for various applications such as image and text retrieval, data filtering and more. With this update also comes a brand new SLURM based inference backend for high-compute environments.

With this users can now:

\ No newline at end of file +
\ No newline at end of file diff --git a/blog/index.html b/blog/index.html index a8ef31a7..81343ccc 100644 --- a/blog/index.html +++ b/blog/index.html @@ -1,4 +1,4 @@ -Blog | LAION

BLOG

Welcome to our LAION blog! Here, you will find commentaries, news, and updates on our current research projects and progress in the field of AI research. These blog posts are not meant to be full scientific research papers, but work in progress to encourage further research / discussions on our discord server and the open scientific community.

Releasing Re-LAION 5B: transparent iteration on LAION-5B with additional safety fixes

by: LAION e.V., 30 Aug, 2024

Today, following a safety revision procedure, we announce Re-LAION-5B, an updated version of LAION-5B, that is the first web-scale, text-link to images pair dataset to be thoroughly cleaned of known links to suspected CSAM. +Blog | LAION

BLOG

Welcome to our LAION blog! Here, you will find commentaries, news, and updates on our current research projects and progress in the field of AI research. These blog posts are not meant to be full scientific research papers, but work in progress to encourage further research / discussions on our discord server and the open scientific community.

Releasing Re-LAION 5B: transparent iteration on LAION-5B with additional safety fixes

by: LAION e.V., 30 Aug, 2024

Today, following a safety revision procedure, we announce Re-LAION-5B, an updated version of LAION-5B, that is the first web-scale, text-link to images pair dataset to be thoroughly cleaned of known links to suspected CSAM. Highlights Re-LAION-5B fixes the issues as reported by Stanford Internet Ob...

BUD-E: Enhancing AI Voice Assistants’ Conversational Quality, Naturalness and Empathy

by: LAION, 08 Feb, 2024

AI voice assistants have revolutionized our interaction with technology, answering queries, performing tasks, and making life easier. However, the stilted, mechanical nature of their responses is a barrier to truly immersive conversational experiences. Unlike human conversation partners, they often ...

LAION POP: 600,000 high-resolution images with detailed descriptions

by: Christoph Schuhmann, Peter Bevan, 17 Nov, 2023

LAION POP is a subset of LAION 5B: This subset comprises 600,000 high-resolution images, each equipped with detailed descriptions. The selection of images was based on 10,000 different concepts popular on the image generation site &quot;Midjourney&quot;. @@ -52,4 +52,4 @@ Large image-text models like ALIGN, BASIC, Turing Bletchl...

LAION-400-MILLION OPEN DATASET

by: Christoph Schuhmann, 20 Aug, 2021

We present LAION-400M: 400M English (image, text) pairs - see also our Data Centric AI NeurIPS Workshop 2021 paper Concept and Content The LAION-400M dataset is entirely openly, freely accessible. -WARNING: be aware that this large-scale dataset is non-curated. It was built for research purposes to e...

\ No newline at end of file +WARNING: be aware that this large-scale dataset is non-curated. It was built for research purposes to e...

\ No newline at end of file diff --git a/blog/laion-400-open-dataset/index.html b/blog/laion-400-open-dataset/index.html index 4624e59f..ec76c459 100644 --- a/blog/laion-400-open-dataset/index.html +++ b/blog/laion-400-open-dataset/index.html @@ -1,4 +1,4 @@ -LAION-400-MILLION OPEN DATASET | LAION

LAION-400-MILLION OPEN DATASET

by: Christoph Schuhmann, 20 Aug, 2021

We present LAION-400M: 400M English (image, text) pairs - see also our Data Centric AI NeurIPS Workshop 2021 paper

+LAION-400-MILLION OPEN DATASET | LAION

LAION-400-MILLION OPEN DATASET

by: Christoph Schuhmann, 20 Aug, 2021

We present LAION-400M: 400M English (image, text) pairs - see also our Data Centric AI NeurIPS Workshop 2021 paper

Concept and Content

The LAION-400M dataset is entirely openly, freely accessible.

WARNING: be aware that this large-scale dataset is non-curated. It was built for research purposes to enable testing model training on larger scale for broad researcher and other interested communities, and is not meant for any real-world production or application.

@@ -185,4 +185,4 @@

LAION-5B: A NEW ERA OF OPEN LARGE-SCALE MULTI-MODAL DATASETS

by: Romain Beaumont, 31 Mar, 2022

We present a dataset of 5,85 billion CLIP-filtered image-text pairs, 14x bigger than LAION-400M, previously the biggest openly accessible image-text dataset in the world - see also our NeurIPS2022 paper

+LAION-5B: A NEW ERA OF OPEN LARGE-SCALE MULTI-MODAL DATASETS | LAION

LAION-5B: A NEW ERA OF OPEN LARGE-SCALE MULTI-MODAL DATASETS

by: Romain Beaumont, 31 Mar, 2022

We present a dataset of 5,85 billion CLIP-filtered image-text pairs, 14x bigger than LAION-400M, previously the biggest openly accessible image-text dataset in the world - see also our NeurIPS2022 paper

See our update on the LAION-5B dataset.

Large image-text models like ALIGN, BASIC, Turing Bletchly, FLORENCE & GLIDE have shown better and better performance compared to previous flagship models like CLIP and DALL-E. Most of them had been trained on billions of image-text pairs and unfortunately, no datasets of this size had been openly available until now. To address this problem we present LAION 5B, a large-scale dataset for research purposes consisting of 5,85B CLIP-filtered image-text pairs. 2,3B contain English language, 2,2B samples from 100+ other languages and 1B samples have texts that do not allow a certain language assignment (e.g. names ). Additionally, we provide several nearest neighbor indices, an improved web interface for exploration & subset creation as well as detection scores for watermark and NSFW. We also announce a full reproduction of a clip training trained on LAION-400M at open_clip. Explore the dataset at the search demo. See also the same post on laion website .

We thank our sponsors hugging face, doodlebot and stability for providing us with computing resources to produce this dataset! We also thank the-eye.eu for hosting the image embeddings and a copy of the whole dataset.

@@ -213,4 +213,4 @@


-

\ No newline at end of file +
\ No newline at end of file diff --git a/blog/laion-aesthetics/index.html b/blog/laion-aesthetics/index.html index ff372ef8..6d70044b 100644 --- a/blog/laion-aesthetics/index.html +++ b/blog/laion-aesthetics/index.html @@ -1,7 +1,7 @@ LAION-Aesthetics | LAION

LAION

LAION-AESTHETICS

by: Christoph Schuhmann, 16 Aug, 2022

We present LAION-Aesthetics, several collections of subsets from LAION 5B with high visual quality.

+<p><img src="https://raw.githubusercontent.com/LAI..."/>

LAION-AESTHETICS

by: Christoph Schuhmann, 16 Aug, 2022

We present LAION-Aesthetics, several collections of subsets from LAION 5B with high visual quality.

To create LAION-Aesthetics we trained several lightweight models that predict the rating people gave when they were asked “How much do you like this image on a scale from 1 to 10?”.

LAION-Aesthetics V1

@@ -45,4 +45,4 @@

join our Discord community and contact us.

Christoph Schuhmann ( spirit-from-germany#1488 ) and Romain Beaumont ( rom1504#5008 )

-
\ No newline at end of file +
\ No newline at end of file diff --git a/blog/laion-coco/index.html b/blog/laion-coco/index.html index e22e174f..6b333840 100644 --- a/blog/laion-coco/index.html +++ b/blog/laion-coco/index.html @@ -1,4 +1,4 @@ -Laion coco: 600M synthetic captions from Laion2B-en | LAION

LAION

LAION COCO: 600M SYNTHETIC CAPTIONS FROM LAION2B-EN

by: Christoph Schuhmann, Andreas Köpf, Richard Vencu, Theo Coombes, Romain Beaumont, 15 Sep, 2022

Author: Christoph Schuhmann, Andreas Köpf , Theo Coombes, Richard Vencu, Benjamin Trom , Romain Beaumont

+Laion coco: 600M synthetic captions from Laion2B-en | LAION

LAION COCO: 600M SYNTHETIC CAPTIONS FROM LAION2B-EN

by: Christoph Schuhmann, Andreas Köpf, Richard Vencu, Theo Coombes, Romain Beaumont, 15 Sep, 2022

Author: Christoph Schuhmann, Andreas Köpf , Theo Coombes, Richard Vencu, Benjamin Trom , Romain Beaumont

We present LAION-COCO, the world’s largest dataset of 600M generated high-quality captions for publicly available web-images

Laion5B has five billion natural captions. They provide a lot of information, but could synthetic captions complement them ?

To answer this question, we use a combination of existing, publicly available models to produce high quality captions for images in the style of MS COCO.

@@ -99,4 +99,4 @@

Benjamin Trom wrote code that help us to convert the .json files to parquet

We thank stability.ai for providing the compute used to generate the captions in the dataset.

-

\ No newline at end of file +
\ No newline at end of file diff --git a/blog/laion-pop/index.html b/blog/laion-pop/index.html index 2f6954ba..5ada725c 100644 --- a/blog/laion-pop/index.html +++ b/blog/laion-pop/index.html @@ -1,4 +1,4 @@ -LAION POP: 600,000 high-resolution images with detailed descriptions | LAION

LAION

LAION POP: 600,000 HIGH-RESOLUTION IMAGES WITH DETAILED DESCRIPTIONS

by: Christoph Schuhmann, Peter Bevan, 17 Nov, 2023

LAION POP is a subset of LAION 5B: This subset comprises 600,000 high-resolution images, each equipped with detailed descriptions. The selection of images was based on 10,000 different concepts popular on the image generation site "Midjourney".

+LAION POP: 600,000 high-resolution images with detailed descriptions | LAION

LAION POP: 600,000 HIGH-RESOLUTION IMAGES WITH DETAILED DESCRIPTIONS

by: Christoph Schuhmann, Peter Bevan, 17 Nov, 2023

LAION POP is a subset of LAION 5B: This subset comprises 600,000 high-resolution images, each equipped with detailed descriptions. The selection of images was based on 10,000 different concepts popular on the image generation site "Midjourney".

@@ -21,4 +21,4 @@

Future Application and Improvements

Although no text-to-image model has been tuned with these data so far, we expect that the use of these data could significantly improve the aesthetic quality of the outputs.

- \ No newline at end of file + \ No newline at end of file diff --git a/blog/laion-stable-horde/index.html b/blog/laion-stable-horde/index.html index e1e27ec0..3776ca83 100644 --- a/blog/laion-stable-horde/index.html +++ b/blog/laion-stable-horde/index.html @@ -1,7 +1,7 @@ Collaboration between LAION and the Stable Horde | LAION

COLLABORATION BETWEEN LAION AND THE STABLE HORDE

by: Konstantinos Thoukydidis, hlky, 08 Jan, 2023

Author: Konstantinos Thoukydidis, hlky

+<p>We are happy to announce that LAIO..."/>

COLLABORATION BETWEEN LAION AND THE STABLE HORDE

by: Konstantinos Thoukydidis, hlky, 08 Jan, 2023

Author: Konstantinos Thoukydidis, hlky

We are happy to announce that LAION will be assisted by the Stable Horde to provide aesthetic ratings for existing datasets and a completely new dataset of Stable Diffusion generations, which will also be rated by their community.

We wrote in the past about LAION-Aesthetics and how we filtered LAION-5b using an aesthetic predictor. The predictor, a simple neural net that uses CLIP ViT-L/14 embeddings as input. hlky has retrained the aesthetic predictor using ViT-H-14, and the results are promising.

hlky’s ViT-H aesthetic predictor was trained on the same datasets as the original - AVA: A Large-Scale Database for Aesthetic Visual Analysis, Simulacra Aesthetic Captions, and LAION-logos. These datasets are limited, totalling around 400k image-rating pairs. This is where the Stable Horde comes in.

@@ -16,4 +16,4 @@

Support this endeavour

If you have any questions or need support about the Stable Horde or the rating system, they have a very active discord server you can join. If you want to support this initiative, you can help by either rating images yourself, or by onboarding your own GPU as a horde worker which will help the community generate more images and thus increase the size of the open dataset they provide.

-

\ No newline at end of file +
\ No newline at end of file diff --git a/blog/laion-translated/index.html b/blog/laion-translated/index.html index f055c9cf..78ffe315 100644 --- a/blog/laion-translated/index.html +++ b/blog/laion-translated/index.html @@ -1,4 +1,4 @@ -Laion translated: 3B captions translated to English from laion5B | LAION

LAION TRANSLATED: 3B CAPTIONS TRANSLATED TO ENGLISH FROM LAION5B

by: Marianna Nezhurina, Romain Beaumont, Richard Vencu and Christoph Schuhmann, 15 Sep, 2022

Author: Marianna Nezhurina Romain Beaumont Richard Vencu Christoph Schuhmann

+Laion translated: 3B captions translated to English from laion5B | LAION

LAION TRANSLATED: 3B CAPTIONS TRANSLATED TO ENGLISH FROM LAION5B

by: Marianna Nezhurina, Romain Beaumont, Richard Vencu and Christoph Schuhmann, 15 Sep, 2022

Author: Marianna Nezhurina Romain Beaumont Richard Vencu Christoph Schuhmann

Laion5B dataset was automatically collected from a section of the human web (common crawl). Can models generate different and interesting data compared to what humans write?

That’s a question we are interested in investigating. To let the community study it, we translated 3B samples of Laion5B from many languages into English.

We released 3 billions captions for the multilingual part of Laion5B. This makes it possible to use the whole Laion5B dataset to train English models. This also enables training models using these aligned pairs such as Multilingual-CLIP.

@@ -117,4 +117,4 @@

https://stability.ai/ for providing the compute for this massive translation. This was a great use of pre-emptible jobs to fill any idle compute available!

-

\ No newline at end of file +
\ No newline at end of file diff --git a/blog/large-openclip/index.html b/blog/large-openclip/index.html index febfb2f4..9fb6fdcd 100644 --- a/blog/large-openclip/index.html +++ b/blog/large-openclip/index.html @@ -1,4 +1,4 @@ -Large scale openCLIP: L/14, H/14 and g/14 trained on LAION-2B | LAION

LARGE SCALE OPENCLIP: L/14, H/14 AND G/14 TRAINED ON LAION-2B

by: Romain Beaumont, 15 Sep, 2022

We trained three large CLIP models with OpenCLIP: ViT-L/14, ViT-H/14 and ViT-g/14 (ViT-g/14 was trained only for about a third the epochs compared to the rest). The H/14 model achieves 78.0% zero shot top-1 accuracy on ImageNet and 73.4% on zero-shot image retrieval at Recall@5 on MS COCO. As of September 2022, this is the best open source CLIP model.

+Large scale openCLIP: L/14, H/14 and g/14 trained on LAION-2B | LAION

LARGE SCALE OPENCLIP: L/14, H/14 AND G/14 TRAINED ON LAION-2B

by: Romain Beaumont, 15 Sep, 2022

We trained three large CLIP models with OpenCLIP: ViT-L/14, ViT-H/14 and ViT-g/14 (ViT-g/14 was trained only for about a third the epochs compared to the rest). The H/14 model achieves 78.0% zero shot top-1 accuracy on ImageNet and 73.4% on zero-shot image retrieval at Recall@5 on MS COCO. As of September 2022, this is the best open source CLIP model.

CLIP makes it possible to compute representations of images and texts to measure how similar they are. It can be used for

  • Zero shot classification: compare an image with the text of the class to know which class is most similar (e.g., ImageNet classification)
    • @@ -283,4 +283,4 @@

    Emad (Stability AI) for providing the many GPUs used during these experiments! (g/14 and H/14!)

    For the L/14 training, we gratefully acknowledge the Gauss Centre for Supercomputing e.V. (www.gauss-centre.eu) for funding this part of work by providing computing time through the John von Neumann Institute for Computing (NIC) on the GCS Supercomputer JUWELS Booster at Jülich Supercomputing Centre (JSC), Germany.

    -

\ No newline at end of file +
\ No newline at end of file diff --git a/blog/leo-lm/index.html b/blog/leo-lm/index.html index 0eaccbc3..cfce4203 100644 --- a/blog/leo-lm/index.html +++ b/blog/leo-lm/index.html @@ -1,4 +1,4 @@ -LeoLM: Igniting German-Language LLM Research | LAION

LEOLM: IGNITING GERMAN-LANGUAGE LLM RESEARCH

by: Björn Plüster, 28 Sep, 2023

We proudly introduce LeoLM (Linguistically Enhanced Open Language Model), the first comprehensive suite of German-language Foundation Language Models trained in collaboration with HessianAI on their new supercomputer 42! Built on Llama-2 and trained on a large-scale, high-quality German text corpus, we present LeoLM-7B and 13B, with LeoLM-70B on the horizon, accompanied by a collection of exceptionally proficient German and bilingual chat models.

+LeoLM: Igniting German-Language LLM Research | LAION

LEOLM: IGNITING GERMAN-LANGUAGE LLM RESEARCH

by: Björn Plüster, 28 Sep, 2023

We proudly introduce LeoLM (Linguistically Enhanced Open Language Model), the first comprehensive suite of German-language Foundation Language Models trained in collaboration with HessianAI on their new supercomputer 42! Built on Llama-2 and trained on a large-scale, high-quality German text corpus, we present LeoLM-7B and 13B, with LeoLM-70B on the horizon, accompanied by a collection of exceptionally proficient German and bilingual chat models.

Meet LeoLM, the first open and commercially available German Foundation Language Model built on Llama-2. Our models extend Llama-2's capabilities into German through continued pretraining on a large corpus of high-quality German and mostly locality-specific text. Thanks to a compute grant at HessianAI's new supercomputer 42, we release two foundation models trained with 8k context length, @@ -95,4 +95,4 @@

OBJAVERSE-XL: AN OPEN DATASET OF OVER 10 MILLION 3D OBJECTS

by: Matt Deitke, 11 Jul, 2023

+<p>We are thrilled to announce Objaverse-XL, an open dataset of over 10 million 3D objects! Using it,..."/>

OBJAVERSE-XL: AN OPEN DATASET OF OVER 10 MILLION 3D OBJECTS

by: Matt Deitke, 11 Jul, 2023

We are thrilled to announce Objaverse-XL, an open dataset of over 10 million 3D objects! Using it, we train Zero123-XL, a foundation model for 3D that displays remarkable generalization abilities. In the landscape of AI, scale has been paramount to recent advances. Over the past decade, we have observed an escalating trend of leveraging large volumes of data to train machine learning models, particularly in NLP and 2D vision. But what about 3D vision tasks? Despite the burgeoning demand for augmented reality (AR) and virtual reality (VR) applications, advancements in 3D vision have lagged, primarily due to the scarcity of high-quality 3D data.

Objaverse 1.0, released back in December, was a step in the right direction, and enabled exciting research like Zero-1-to-3 for novel view synthesis and single view 3D reconstruction. But, it was still quite small, being on the order of 800K objects. With Objaverse-XL, we scale up the number of 3D objects that we use from 800K to over 10 million deduplicated 3D objects, pulling in objects from a variety of sources.

@@ -24,4 +24,4 @@

-

\ No newline at end of file +
\ No newline at end of file diff --git a/blog/oig-dataset/index.html b/blog/oig-dataset/index.html index ce5941fb..a400d5ea 100644 --- a/blog/oig-dataset/index.html +++ b/blog/oig-dataset/index.html @@ -1,4 +1,4 @@ -The OIG Dataset | LAION

LAION

THE OIG DATASET

by: By Huu Nguyen - Ontocord.ai, Sameer Suri, Ken Tsui , Shahules786, Together.xyz team, and Christoph Schuhmann - LAION.ai, 10 Mar, 2023

The Open Instruction Generalist (OIG) dataset is a large open source instruction dataset that currently contains ~43M instructions.

+The OIG Dataset | LAION

THE OIG DATASET

by: By Huu Nguyen - Ontocord.ai, Sameer Suri, Ken Tsui , Shahules786, Together.xyz team, and Christoph Schuhmann - LAION.ai, 10 Mar, 2023

The Open Instruction Generalist (OIG) dataset is a large open source instruction dataset that currently contains ~43M instructions.

OIG is one of many chatbot datasets that LAION, along with its volunteers, Ontocord, Together and other members of the open source community, will be releasing and is intended to create equal access to chatbot technology. Everyone is welcome to use the dataset and contribute improvements to it.

Examples of what is in OIG

@@ -184,4 +184,4 @@

Examples from the Com2Sense and Strategy QA datasets that were reformatted into natural instructions using large language models with few shot prompting and additional quality filtering steps.

Character and Scene Descriptions

Examples of instructions and responses for the generation of character or scene descriptions. Scenes were sourced from video game wikis and reformatted into instruction / response format using large language models or generated by few shot prompting with large language models.

-

\ No newline at end of file +
\ No newline at end of file diff --git a/blog/open-empathic/index.html b/blog/open-empathic/index.html index ae98e84d..937128fe 100644 --- a/blog/open-empathic/index.html +++ b/blog/open-empathic/index.html @@ -1,7 +1,7 @@ Open Empathic Launch | LAION

LAION

OPEN EMPATHIC LAUNCH

by: Christoph, Knoriy, Robert, 22 Oct, 2023

We are thrilled to present Open Empathic, a pioneering open-source project initiated by our non-profit organization, LAION.

+<p>Open E..."/>

OPEN EMPATHIC LAUNCH

by: Christoph, Knoriy, Robert, 22 Oct, 2023

We are thrilled to present Open Empathic, a pioneering open-source project initiated by our non-profit organization, LAION.

Open Empathic aims to equip open-source AI systems with empathy and emotional intelligence. We hope that methods and tools developed within the framework of this project, together with a community of researchers and technology enthusiasts, will revolutionize the way AI interacts with and supports humans in various domains.

In an increasingly AI-driven world, it is of paramount importance that AI systems possess emotional intelligence to understand and respond to human emotions. As AI plays an ever-expanding role in our daily lives, ranging from education to healthcare, elderly care, and commercial contexts, it becomes vital to prioritize the well-being and emotional intelligence of AI-human interactions.

@@ -41,4 +41,4 @@

Join the Open Empathic movement today, and let's shape a compassionate, empathic AI future together!

Sincerely,

The LAION Team

- \ No newline at end of file + \ No newline at end of file diff --git a/blog/open-flamingo-v2/index.html b/blog/open-flamingo-v2/index.html index ff465c5d..1b266431 100644 --- a/blog/open-flamingo-v2/index.html +++ b/blog/open-flamingo-v2/index.html @@ -1,4 +1,4 @@ -OpenFlamingo v2: New Models and Enhanced Training Setup | LAION

OPENFLAMINGO V2: NEW MODELS AND ENHANCED TRAINING SETUP

by: Anas Awadalla* and Irena Gao*, 28 Jun, 2023

[GitHub] [Demo] [Models]

+OpenFlamingo v2: New Models and Enhanced Training Setup | LAION

OPENFLAMINGO V2: NEW MODELS AND ENHANCED TRAINING SETUP

by: Anas Awadalla* and Irena Gao*, 28 Jun, 2023

[GitHub] [Demo] [Models]

About three months ago, we announced OpenFlamingo, an open-source effort to replicate DeepMind's Flamingo models.

Today, we are excited to release five trained OpenFlamingo models across the 3B, 4B, and 9B scales. These models are based on Mosaic’s MPT-1B and 7B and Together.xyz’s RedPajama-3B, meaning they are built on open-source models with less restrictive licenses than LLaMA. When averaging performance across 7 evaluation datasets, OpenFlamingo models attain more than 80% of the performance of their corresponding Flamingo model. OpenFlamingo-3B and OpenFlamingo-9B also attain more than 60% of fine-tuned SOTA performance using just 32 in-context examples.

@@ -68,4 +68,4 @@

Josh Gardner, Jack Hessel, Yusuf Hanafy, Wanrong Zhu, Kalyani Marathe, Yonatan Bitton, Samir Gadre, Shiori Sagawa, Jenia Jitsev, Simon Kornblith, Pang Wei Koh, Gabriel Ilharco, Mitchell Wortsman, and Ludwig Schmidt

Acknowledgements

We would like to thank Jean-Baptiste Alayrac and Antoine Miech for their advice and Stability AI for providing us with compute resources to train these models.

-
\ No newline at end of file +
\ No newline at end of file diff --git a/blog/open-flamingo/index.html b/blog/open-flamingo/index.html index 268f96a0..f839bbf3 100644 --- a/blog/open-flamingo/index.html +++ b/blog/open-flamingo/index.html @@ -1,7 +1,7 @@ Announcing OpenFlamingo: An open-source framework for training vision-language models with in-context learning | LAION

ANNOUNCING OPENFLAMINGO: AN OPEN-SOURCE FRAMEWORK FOR TRAINING VISION-LANGUAGE MODELS WITH IN-CONTEXT LEARNING

by: Anas Awadalla and Irena Gao, 28 Mar, 2023

Overview. +We are thrilled to announce the release of OpenFlamingo, an open-source reproduction of DeepMind's Flamingo model. At its core,..."/>

ANNOUNCING OPENFLAMINGO: AN OPEN-SOURCE FRAMEWORK FOR TRAINING VISION-LANGUAGE MODELS WITH IN-CONTEXT LEARNING

by: Anas Awadalla and Irena Gao, 28 Mar, 2023

Overview. We are thrilled to announce the release of OpenFlamingo, an open-source reproduction of DeepMind's Flamingo model. At its core, OpenFlamingo is a framework that enables training and evaluation of large multimodal models (LMMs). Check out our GitHub repository and demo to get started!

For this first release, our contributions are as follows:

\ No newline at end of file +
\ No newline at end of file diff --git a/blog/open-lm/index.html b/blog/open-lm/index.html index 6df75d0a..bc8ac0b8 100644 --- a/blog/open-lm/index.html +++ b/blog/open-lm/index.html @@ -1,7 +1,7 @@ Introducing OpenLM | LAION

INTRODUCING OPENLM

by: OpenLM team, 26 Sep, 2023

+<h2><a id="introduction" class="anchor" href="#introduction" aria-hidden="true"><svg aria-hidde..."/>

INTRODUCING OPENLM

by: OpenLM team, 26 Sep, 2023

Introduction

We release OpenLM a simple and minimalist PyTorch codebase for training medium-sized language models. OpenLM is designed to maximize GPU utilization and training speed, and is easy to modify for new language model research and applications.

We validate OpenLM by training two language models, OpenLM-1B and OpenLM-7B, on 1.6T and 1.25T tokens of text, respectively. We evaluate these models on standard zero-shot text classification and multiple choice tasks and find that OpenLM-1B outperforms many popular, similarly sized models such as OPT-1.3B and Pythia-1B. OpenLM-7B achieves similar performance to LLAMA-7B and MPT-7B.

@@ -571,4 +571,4 @@

open-clip developed by a team including Ross Wightman, Romain Beaumont, Cade Gordon, Mehdi Cherti, Jenia Jitsev, and open-flamingo, developed by a team including Anas Awadalla and Irena Gao. Additional inspiration is from lit-llama.

We thank Stability AI for providing the compute for this project, the RedPajama team for their dataset, Sarah Pratt for logo design, IFML, and Toyota Research Institute. We also thank the following people for helpful advice and feedback throughout the project: Jonathan Frankle, Daniel King, Luca Soldaini.

-

\ No newline at end of file +
\ No newline at end of file diff --git a/blog/paella/index.html b/blog/paella/index.html index 757dd238..1187ffb7 100644 --- a/blog/paella/index.html +++ b/blog/paella/index.html @@ -1,7 +1,7 @@ A new Paella: Simple & Efficient Text-To-Image generation | LAION

A NEW PAELLA: SIMPLE & EFFICIENT TEXT-TO-IMAGE GENERATION

by: Dominic Rampas and Pablo Pernias, 15 Apr, 2023

+<h3><a id="overview" class="a..."/>

A NEW PAELLA: SIMPLE & EFFICIENT TEXT-TO-IMAGE GENERATION

by: Dominic Rampas and Pablo Pernias, 15 Apr, 2023

Overview.

We are releasing a new Paella model which builds on top of our initial paper https://arxiv.org/abs/2211.07292. Paella is a text-to-image model that works in a quantized latent space and learns similarly to MUSE and Diffusion models. @@ -111,4 +111,4 @@

Richard Vencu for an incredible amount of help regarding hardware issues.
  • StabilityAI for providing GPU-Cluster access and faith in Paella.
    • -

    \ No newline at end of file +
    \ No newline at end of file diff --git a/blog/petition/index.html b/blog/petition/index.html index 66976272..e3730860 100644 --- a/blog/petition/index.html +++ b/blog/petition/index.html @@ -1,4 +1,4 @@ -Petition for keeping up the progress tempo on AI research while securing its transparency and safety. | LAION

    PETITION FOR KEEPING UP THE PROGRESS TEMPO ON AI RESEARCH WHILE SECURING ITS TRANSPARENCY AND SAFETY.

    by: LAION.ai, 29 Mar, 2023

    LINK TO OUR PETITION

    +Petition for keeping up the progress tempo on AI research while securing its transparency and safety. | LAION

    PETITION FOR KEEPING UP THE PROGRESS TEMPO ON AI RESEARCH WHILE SECURING ITS TRANSPARENCY AND SAFETY.

    by: LAION.ai, 29 Mar, 2023

    LINK TO OUR PETITION

    Authors: Christoph Schuhmann, Huu Nguyen, Robert Kaczmarczyk, Jenia Jitsev & LAION community

    Securing Our Digital Future: Calling for CERN like international organization to transparently coordinate and progress on large-scale AI research and its safety

    In an era of unparalleled technological advancements, humanity stands on the precipice of a new epoch characterized by the profound influence of artificial intelligence (AI) and its foundational models, such as GPT-4. The potential applications of these technologies are vast, spanning scientific research, education, governance, and small and medium-sized enterprises. To harness their full potential as tools for societal betterment, it is vital to democratize research on and access to them, lest we face severe repercussions for our collective future.

    @@ -18,4 +18,4 @@

    What you can do

    We urge you to join us in this crucial campaign. Sign this petition and make your voice heard. Our collective digital future, the autonomy of our academic research, and the equilibrium of our global economy depend on our ability to act quickly and decisively. Together, we can build a future where advanced AI technologies are accessible to all, and where innovation and progress are not constrained by the boundaries of a few powerful corporations. Let us seize this opportunity and build a brighter future for generations to come.

    -
    \ No newline at end of file +
    \ No newline at end of file diff --git a/blog/relaion-5b/index.html b/blog/relaion-5b/index.html index 4dcec931..658f9b23 100644 --- a/blog/relaion-5b/index.html +++ b/blog/relaion-5b/index.html @@ -1,4 +1,4 @@ -Releasing Re-LAION 5B: transparent iteration on LAION-5B with additional safety fixes | LAION

    RELEASING RE-LAION 5B: TRANSPARENT ITERATION ON LAION-5B WITH ADDITIONAL SAFETY FIXES

    by: LAION e.V., 30 Aug, 2024

    Today, following a safety revision procedure, we announce Re-LAION-5B, an updated version of LAION-5B, that is the first web-scale, text-link to images pair dataset to be thoroughly cleaned of known links to suspected CSAM.

    +Releasing Re-LAION 5B: transparent iteration on LAION-5B with additional safety fixes | LAION

    RELEASING RE-LAION 5B: TRANSPARENT ITERATION ON LAION-5B WITH ADDITIONAL SAFETY FIXES

    by: LAION e.V., 30 Aug, 2024

    Today, following a safety revision procedure, we announce Re-LAION-5B, an updated version of LAION-5B, that is the first web-scale, text-link to images pair dataset to be thoroughly cleaned of known links to suspected CSAM.

    Highlights

    • Re-LAION-5B fixes the issues as reported by Stanford Internet Observatory in December 2023 for the original LAION-5B and is available for download in two versions, Re-LAION-5B research and Re-LAION-5B research-safe. The work was completed in partnership with the Internet Watch Foundation (IWF), the Canadian Center for Child Protection (C3P), and Stanford Internet Observatory. For the work, we utilized lists of link and image hashes provided by our partners, as of July 2024.
      • @@ -110,4 +110,4 @@

      LEGAL DISCLAIMER

      The datasets of LAION only contain links and metadata. LAION is not responsible for the content that can be accessed via the links. LAION researchers do not inspect the content of individual samples either, relying on overall statistics collected across all samples, and the filtering is automated due to the huge amount of data. LAION has never distributed image content itself.

      LAION has been committed to removing illegal content from its datasets from the very beginning (see original announcement from 20.08.2021) and has implemented appropriate measures to achieve this from the outset. LAION strictly adheres to the principle that illegal content is removed ASAP after it becomes known.

      -
    \ No newline at end of file +
    \ No newline at end of file diff --git a/blog/strategic-game-dataset/index.html b/blog/strategic-game-dataset/index.html index a0c8663c..07409809 100644 --- a/blog/strategic-game-dataset/index.html +++ b/blog/strategic-game-dataset/index.html @@ -1,4 +1,4 @@ -Strategic Game Datasets for Enhancing AI Planning: An Invitation for Collaborative Research | LAION

    STRATEGIC GAME DATASETS FOR ENHANCING AI PLANNING: AN INVITATION FOR COLLABORATIVE RESEARCH

    by: Christoph Schuhmann & Qi Sun, 18 Oct, 2023

    Recent advancements in artificial intelligence (AI) underscore the progress of reasoning and planning shown by recent generalist machine learning (ML) models. The progress can be boosted by datasets that can further boost these generic capabilities when used for training foundation models of various kind. This research initiative has generated extensive synthetic datasets from complex games — chess, Rubik's Cube, and mazes — to study facilitation and the advancement of these critical generic skills in AI models. This paper delineates the methodology, dataset structure, and preliminary analysis, concluding with an open invitation for collaborative research.

    +Strategic Game Datasets for Enhancing AI Planning: An Invitation for Collaborative Research | LAION

    STRATEGIC GAME DATASETS FOR ENHANCING AI PLANNING: AN INVITATION FOR COLLABORATIVE RESEARCH

    by: Christoph Schuhmann & Qi Sun, 18 Oct, 2023

    Recent advancements in artificial intelligence (AI) underscore the progress of reasoning and planning shown by recent generalist machine learning (ML) models. The progress can be boosted by datasets that can further boost these generic capabilities when used for training foundation models of various kind. This research initiative has generated extensive synthetic datasets from complex games — chess, Rubik's Cube, and mazes — to study facilitation and the advancement of these critical generic skills in AI models. This paper delineates the methodology, dataset structure, and preliminary analysis, concluding with an open invitation for collaborative research.

    Introduction

    The field of AI has observed a pivotal shift toward foundation generalist models capable of advanced strategic planning, essential for complex problem-solving tasks. Recognizing the potential of various complex games as ideal proxies for real-world problems, this research focuses on the generation of large-scale synthetic datasets. These datasets are designed to challenge and enhance the strategic planning capabilities of generative pre-trained transformers (GPT) and similar models.

    Dataset Overview

    @@ -18,4 +18,4 @@

    Conclusively, this initiative marks a significant stride toward intricate problem-solving and strategic planning in AI, extending an open invitation to the research community for collaborative advancement in this domain.

    Acknowledgements

    Special thanks to Prof. Rio Yokota for providing the necessary compute time on the Fugaku supercomputer and Yago Kastro for coding the first draft of the chess selfplay script.

    -

    \ No newline at end of file +
    \ No newline at end of file diff --git a/blog/transparent-ai/index.html b/blog/transparent-ai/index.html index 7dc09abb..8a753134 100644 --- a/blog/transparent-ai/index.html +++ b/blog/transparent-ai/index.html @@ -1,4 +1,4 @@ -Towards a transparent AI Future: The Call for less regulatory hurdles on Open-Source AI in Europe | LAION

    TOWARDS A TRANSPARENT AI FUTURE: THE CALL FOR LESS REGULATORY HURDLES ON OPEN-SOURCE AI IN EUROPE

    by: LAION, 21 Sep, 2023

    Following our previous open letter to the European Parliament on the significance of open-source AI, LAION, backed by European Laboratory for Learning and Intelligent Systems (ELLIS) and a long list of very impactful AI researchers, we submit this new open letter to the European Parliament:

    +Towards a transparent AI Future: The Call for less regulatory hurdles on Open-Source AI in Europe | LAION

    TOWARDS A TRANSPARENT AI FUTURE: THE CALL FOR LESS REGULATORY HURDLES ON OPEN-SOURCE AI IN EUROPE

    by: LAION, 21 Sep, 2023

    Following our previous open letter to the European Parliament on the significance of open-source AI, LAION, backed by European Laboratory for Learning and Intelligent Systems (ELLIS) and a long list of very impactful AI researchers, we submit this new open letter to the European Parliament:

    @@ -119,4 +119,4 @@

    VIDEO2DATASET: A SIMPLE TOOL FOR LARGE VIDEO DATASET CURATION

    by: Maciej Kilian, 10 Jul, 2023

    [GitHub]

    +<p>Within only two years large foundational models like <a href="https://arxiv.org/abs/..."/>

    VIDEO2DATASET: A SIMPLE TOOL FOR LARGE VIDEO DATASET CURATION

    by: Maciej Kilian, 10 Jul, 2023

    [GitHub]

    Within only two years large foundational models like CLIP, Stable Diffusion, and Flamingo have fundamentally transformed multimodal deep learning. Because of such models and their impressive capabilities to either create stunning, high-resolution imagery or to solve complex downstream tasks, joint text-image modeling has emerged from a niche application to one of the (or maybe the) most relevant topics in today’s AI landscape. Remarkably, all these models, despite addressing very different tasks and being very different in design, share three fundamental properties as the main drivers behind their strong performance: A simple and stable objective function during (pre-)training, a well-investigated scalable model architecture, and - probably most importantly - a large diverse dataset.

    As of 2023, multimodal deep learning is still heavily focusing on text-image modeling, while other modalities such as video (and audio) are only sparsely investigated. Since the algorithms to train the above models are usually modality agnostic, one might wonder why there aren’t strong foundational models for these additional modalities. The reason for this is – plain and simple – the lacking availability of large scale, annotated datasets. As opposed to image modeling, where there are established datasets for scaling such as LAION-5B, DataComp, and COYO-700M and scalable tools as img2dataset, this lack of clean data hinders research and development of large multimodal models especially for the video domain.

    We argue that overcoming this data problem is a core interest of (open source) multimodal research since it can foster important previously impossible projects such as high quality video and audio generation, better pre-trained models for robotics, movie AD for the blind community, and more.

    @@ -128,4 +128,4 @@

    Andreas for greatly improving the video2dataset dataloader and implementing slurm distribution.
  • Sumith for implementing synthetic captioning and lots of help during writing the blogpost (especially with visualizations).
    • -

    \ No newline at end of file +
    \ No newline at end of file diff --git a/blog/visit_bench/index.html b/blog/visit_bench/index.html index 33a0f96e..baa4e268 100644 --- a/blog/visit_bench/index.html +++ b/blog/visit_bench/index.html @@ -1,4 +1,4 @@ -Introducing VisIT-Bench, a new benchmark for instruction-following vision-language models inspired by real-world use | LAION

    LAION

    INTRODUCING VISIT-BENCH, A NEW BENCHMARK FOR INSTRUCTION-FOLLOWING VISION-LANGUAGE MODELS INSPIRED BY REAL-WORLD USE

    by: Yonatan Bitton, 15 Aug, 2023

    [Paper] [Code] [Dataset] [Leaderboard]

    +Introducing VisIT-Bench, a new benchmark for instruction-following vision-language models inspired by real-world use | LAION

    INTRODUCING VISIT-BENCH, A NEW BENCHMARK FOR INSTRUCTION-FOLLOWING VISION-LANGUAGE MODELS INSPIRED BY REAL-WORLD USE

    by: Yonatan Bitton, 15 Aug, 2023

    [Paper] [Code] [Dataset] [Leaderboard]

    We are thrilled to introduce VisIT-Bench, a benchmark for evaluating instruction-following vision-language models (VLMs). The central goal of VisIT-Bench is to provide a more accurate and meaningful assessment of VLMs, particularly in the context of human-chatbot interactions inspired by real-world scenarios.

    VisIT-Bench comprises 678 examples. Each example includes:

      @@ -60,4 +60,4 @@

      In Conclusion

      VisIT-Bench offers a comprehensive lens on VLMs by utilizing 70 carefully curated instruction families, mirroring a wide range of real-world scenarios. This approach allows an in-depth assessment of model understanding but paves the way for enhancing VLMs' performance across various tasks. VisIT-Bench is dynamic to participate, practitioners simply submit their model's response on the project website; Data, code and leaderboard is available at the project website.

      -
    \ No newline at end of file +
    \ No newline at end of file diff --git a/dataset-requests/index.html b/dataset-requests/index.html index aec62a92..9ca41e2d 100644 --- a/dataset-requests/index.html +++ b/dataset-requests/index.html @@ -1 +1 @@ -Dataset Requests | LAION

    LAION

    DATASET REQUESTS

    Submitting issues with the dataset

    Please let us know of any problem found in the datasets by submitting to the following form. By doing so, you agree to our privacy policy.

    By submitting to the form, you agree to our privacy policy.
    \ No newline at end of file +Dataset Requests | LAION

    LAION

    DATASET REQUESTS

    Submitting issues with the dataset

    Please let us know of any problem found in the datasets by submitting to the following form. By doing so, you agree to our privacy policy.

    By submitting to the form, you agree to our privacy policy.
    \ No newline at end of file diff --git a/donations/index.html b/donations/index.html index 0551cc97..49487c1c 100644 --- a/donations/index.html +++ b/donations/index.html @@ -1,3 +1,3 @@ -Donations | LAION

    LAION

    DONATIONS

    LAION is a non-profit organizaton solely relying on donations.

    +Donations | LAION

    DONATIONS

    LAION is a non-profit organizaton solely relying on donations.

    If you want to help us democraticising AI research, head over to our gofundme page or donate directly to the following bank account:

    -
    • Name: LAION e.V.
    • IBAN: DE26430609671260733600
    • BIC: GENODEM1GLS
    \ No newline at end of file +
    • Name: LAION e.V.
    • IBAN: DE26430609671260733600
    • BIC: GENODEM1GLS
    \ No newline at end of file diff --git a/faq/index.html b/faq/index.html index acae56ee..6b8a3a8e 100644 --- a/faq/index.html +++ b/faq/index.html @@ -1 +1 @@ -FAQ | LAION

    LAION

    FAQ

    Does LAION datasets respect copyright laws?

    LAION datasets are simply indexes to the internet, i.e. lists of URLs to the original images together with the ALT texts found linked to those images. While we downloaded and calculated CLIP embeddings of the pictures to compute similarity scores between pictures and texts, we subsequently discarded all the photos. Any researcher using the datasets must reconstruct the images data by downloading the subset they are interested in. For this purpose, we suggest the img2dataset tool.

    Do the datasets contain images that may be disturbing for viewers?

    No, but links in the datasets can lead to images that are disturbing or discomforting depending on the filter or search method employed.

    I found a dataset containing images while searching on the internet. What about copyright then?

    Any dataset containing images is not released by LAION, it must have been reconstructed with the provided tools by other people. We do not host and also do not provide links on our website to access such datasets. Please refer only to links we provide for official released data.

    I found my name and my picture in the dataset. I am an EU citizen and I want to protect my personal data as allowed by GDPR. What should I do?

    If you found your name only on the ALT text data, and the corresponding picture does NOT contain your image, this is not considered personal data under GDPR terms. Your name associated with other identifiable data is. If the URL or the picture has your image, you may request a takedown of the dataset entry in the GDPR page. As per GDPR, we provide a takedown form you can use. Upon form submission, we will investigate the request, and if verifiable, we will remove the entry from all data repositories we control. Such repositories include current data stored on our computers and future releases of the datasets. We cannot act on data that are not under our control, for example, past releases that circulate via torrents.

    Do your scripts respect robots.txt instructions?

    Despite the “Crawling at Home” project name, we are not crawling websites to create the datasets. Common Crawl did the crawling part in the past, and they did respect the robots.txt instruction. We only analyse their data and then look at the pictures to assess their value concerning the provided alt text.

    \ No newline at end of file +FAQ | LAION

    LAION

    FAQ

    Does LAION datasets respect copyright laws?

    LAION datasets are simply indexes to the internet, i.e. lists of URLs to the original images together with the ALT texts found linked to those images. While we downloaded and calculated CLIP embeddings of the pictures to compute similarity scores between pictures and texts, we subsequently discarded all the photos. Any researcher using the datasets must reconstruct the images data by downloading the subset they are interested in. For this purpose, we suggest the img2dataset tool.

    Do the datasets contain images that may be disturbing for viewers?

    No, but links in the datasets can lead to images that are disturbing or discomforting depending on the filter or search method employed.

    I found a dataset containing images while searching on the internet. What about copyright then?

    Any dataset containing images is not released by LAION, it must have been reconstructed with the provided tools by other people. We do not host and also do not provide links on our website to access such datasets. Please refer only to links we provide for official released data.

    I found my name and my picture in the dataset. I am an EU citizen and I want to protect my personal data as allowed by GDPR. What should I do?

    If you found your name only on the ALT text data, and the corresponding picture does NOT contain your image, this is not considered personal data under GDPR terms. Your name associated with other identifiable data is. If the URL or the picture has your image, you may request a takedown of the dataset entry in the GDPR page. As per GDPR, we provide a takedown form you can use. Upon form submission, we will investigate the request, and if verifiable, we will remove the entry from all data repositories we control. Such repositories include current data stored on our computers and future releases of the datasets. We cannot act on data that are not under our control, for example, past releases that circulate via torrents.

    Do your scripts respect robots.txt instructions?

    Despite the “Crawling at Home” project name, we are not crawling websites to create the datasets. Common Crawl did the crawling part in the past, and they did respect the robots.txt instruction. We only analyse their data and then look at the pictures to assess their value concerning the provided alt text.

    \ No newline at end of file diff --git a/impressum/index.html b/impressum/index.html index db47b5a2..f1c9c4e1 100644 --- a/impressum/index.html +++ b/impressum/index.html @@ -1,4 +1,4 @@ -Impressum | LAION

    LAION

    IMPRESSUM

    Angaben gemäß § 5 TMG

    +Impressum | LAION

    IMPRESSUM

    Angaben gemäß § 5 TMG

    LAION gemeinnütziger e.V.

    Marlowring 26

    22525 Hamburg

    @@ -25,4 +25,4 @@

    Haftungsausschluss

    Die Nutzung unserer Webseite ist in der Regel ohne Angabe personenbezogener Daten möglich. Soweit auf unseren Seiten personenbezogene Daten (beispielsweise Name, Anschrift oder eMail-Adressen) erhoben werden, erfolgt dies, soweit möglich, stets auf freiwilliger Basis. Diese Daten werden ohne Ihre ausdrückliche Zustimmung nicht an Dritte weitergegeben.

    Wir weisen darauf hin, dass die Datenübertragung im Internet (z.B. bei der Kommunikation per E-Mail) Sicherheitslücken aufweisen kann. Ein lückenloser Schutz der Daten vor dem Zugriff durch Dritte ist nicht möglich.

    Der Nutzung von im Rahmen der Impressumspflicht veröffentlichten Kontaktdaten durch Dritte zur Übersendung von nicht ausdrücklich angeforderter Werbung und Informationsmaterialien wird hiermit ausdrücklich widersprochen. Die Betreiber der Seiten behalten sich ausdrücklich rechtliche Schritte im Falle der unverlangten Zusendung von Werbeinformationen, etwa durch Spam-Mails, vor.

    -
    \ No newline at end of file +
    \ No newline at end of file diff --git a/index.html b/index.html index d490aa69..6ec38330 100644 --- a/index.html +++ b/index.html @@ -1 +1 @@ -LAION

    LAION

    LAION

    Large-scale Artificial Intelligence Open Network

    TRULY 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-400M

    An open dataset containing 400 million English image-text pairs.

    LAION-5B

    A dataset consisting of 5.85 billion multilingual CLIP-filtered image-text pairs.

    Clip H/14

    The largest CLIP (Contrastive Language-Image Pre-training) vision transformer model.

    LAION-Aesthetics

    A subset of LAION-5B filtered by a model trained to score aesthetically pleasing images.

    \ No newline at end of file +LAION

    LAION

    LAION

    Large-scale Artificial Intelligence Open Network

    TRULY 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-400M

    An open dataset containing 400 million English image-text pairs.

    LAION-5B

    A dataset consisting of 5.85 billion multilingual CLIP-filtered image-text pairs.

    Clip H/14

    The largest CLIP (Contrastive Language-Image Pre-training) vision transformer model.

    LAION-Aesthetics

    A subset of LAION-5B filtered by a model trained to score aesthetically pleasing images.

    \ No newline at end of file diff --git a/laion-400-open-dataset/index.html b/laion-400-open-dataset/index.html index c524f31d..9ebd6332 100644 --- a/laion-400-open-dataset/index.html +++ b/laion-400-open-dataset/index.html @@ -1 +1 @@ -

    LAION

    \ No newline at end of file +

    LAION

    \ No newline at end of file diff --git a/laion-5b-a-new-era-of-open-large-scale-multi-modal-datasets/index.html b/laion-5b-a-new-era-of-open-large-scale-multi-modal-datasets/index.html index 04804033..7ff61f4d 100644 --- a/laion-5b-a-new-era-of-open-large-scale-multi-modal-datasets/index.html +++ b/laion-5b-a-new-era-of-open-large-scale-multi-modal-datasets/index.html @@ -1 +1 @@ -

    LAION

    \ No newline at end of file +

    LAION

    \ No newline at end of file diff --git a/notes/cpretrain/index.html b/notes/cpretrain/index.html index 5a1d6445..c699c13e 100644 --- a/notes/cpretrain/index.html +++ b/notes/cpretrain/index.html @@ -1,4 +1,4 @@ -Conditional Pretraining of Large Language Models | LAION

    LAION

    CONDITIONAL PRETRAINING OF LARGE LANGUAGE MODELS

    by: Rallio, 16 May, 2023

    Introduction

    +Conditional Pretraining of Large Language Models | LAION

    CONDITIONAL PRETRAINING OF LARGE LANGUAGE MODELS

    by: Rallio, 16 May, 2023

    Introduction

    Large language models (LLMs), such as OpenAI's ChatGPT and similar chatbot products from other organizations, have recently gained widespread adoption. These models can extend text or respond to instructions in a natural and helpful manner. Despite the core technologies behind LLMs, namely the transformer architecture and the GPT decoder-only causal language model, remaining relatively unchanged for over five years, the surge in popularity of ChatGPT can be largely attributed to recent approaches that better align the output of LLMs with users' and service providers' intentions.

    Two primary approaches have been employed to better align large language models with human expectations. The first is known as supervised finetuning (SFT) on natural instructions, while the second is called reinforcement learning from human feedback (RLHF). Both methods aim to improve the performance and usability of LLMs, but they differ in their implementation. SFT involves training the model using labeled datasets that contain natural instructions, which helps the model understand and respond more accurately to user queries. RLHF, on the other hand, is a technique that uses human preferences as a reward signal to fine-tune models. It involves collecting a dataset of human-written demonstrations on prompts, training supervised learning baselines, and then gathering a dataset of human-labeled comparisons between two model outputs on a larger set of prompts. A reward model (RM) is trained on this dataset to predict which output labelers would prefer, and this RM is used as a reward function to fine-tune the LLM using the PPO algorithm. However, there is an "alignment tax" associated with this approach, which can result in worse performance in some situations.

    cond_pretrain_im1

    @@ -51,4 +51,4 @@

  • PALM-2 Technical Report by Google AI. Search for "control tokens" to find relevant information.
    • -

    \ No newline at end of file +
    \ No newline at end of file diff --git a/notes/general-gpt/index.html b/notes/general-gpt/index.html index ae7fccc2..c1f40a4c 100644 --- a/notes/general-gpt/index.html +++ b/notes/general-gpt/index.html @@ -1,4 +1,4 @@ -General-GPT: Breaking the Modality Constraint | LAION

    LAION

    GENERAL-GPT: BREAKING THE MODALITY CONSTRAINT

    by: Shivaen Ramshetty and Christoph Schuhmann, 28 Mar, 2023

    Introduction

    +General-GPT: Breaking the Modality Constraint | LAION

    GENERAL-GPT: BREAKING THE MODALITY CONSTRAINT

    by: Shivaen Ramshetty and Christoph Schuhmann, 28 Mar, 2023

    Introduction

    With the rapid explosion of large language models and utilization of their encompassing applications, most notably ChatGPT, there is a clear promise of more capable and useful AI models/systems. Often, such models are compared to us as humans using the Turing test or their performance on tasks relative to humans. As of recent, these models have even achieved incredible success on tests designed for humans such as the LSAT. However, the limited means by which one can interact with such systems elucidates a variety of opportunities for exploration and possibly discovery. We ask whether modalities can be mixed and learnt alongside one another, and whether that environment of learning offers new avenues for understanding.

    With this in mind, we are excited to introduce a relatively new project at LAION called General-GPT.

    Goals

    @@ -126,4 +126,4 @@

    NOTES

    Welcome to our LAION notes section! Here, you will find quick overviews or work in progress of the recent research by our community!

    Call to Build Open Multi-Modal Models for Personal Assistants

    by: LAION, 28 Jun, 2024

    We’re pleased to announce the World's first Large Competitive Debate Dataset: LAION-Debate. LAION-Debate is a large Competitive debate dataset providing links to Competitive Debate Championships, discussions and prominent speakers intake and conversations posted on YouTube by University of Cambridge...

    Call to Build Open Multi-Modal Models for Personal Assistants

    by: Christoph Schuhmann, 29 May, 2024

    Technologies like the recently introduced GPT-4-OMNI from OpenAI show again the potential which strong multi-modal models might have to positively transform many aspects of our lives. A particularly impressive example of this is in the field of education. Imagine every person in the world having the...

    Safety Review for LAION 5B

    by: LAION.ai, 19 Dec, 2023

    There have been reports in the press about the results of a research project at Stanford University, according to which the LAION training set 5B contains potentially illegal content in the form of CSAM. We would like to comment on this as follows: +Blog | LAION

    NOTES

    Welcome to our LAION notes section! Here, you will find quick overviews or work in progress of the recent research by our community!

    Call to Build Open Multi-Modal Models for Personal Assistants

    by: LAION, 28 Jun, 2024

    We’re pleased to announce the World's first Large Competitive Debate Dataset: LAION-Debate. LAION-Debate is a large Competitive debate dataset providing links to Competitive Debate Championships, discussions and prominent speakers intake and conversations posted on YouTube by University of Cambridge...

    Call to Build Open Multi-Modal Models for Personal Assistants

    by: Christoph Schuhmann, 29 May, 2024

    Technologies like the recently introduced GPT-4-OMNI from OpenAI show again the potential which strong multi-modal models might have to positively transform many aspects of our lives. A particularly impressive example of this is in the field of education. Imagine every person in the world having the...

    Safety Review for LAION 5B

    by: LAION.ai, 19 Dec, 2023

    There have been reports in the press about the results of a research project at Stanford University, according to which the LAION training set 5B contains potentially illegal content in the form of CSAM. We would like to comment on this as follows: LAION is a non-profit organization that provides da...

    Conditional Pretraining of Large Language Models

    by: Rallio, 16 May, 2023

    Introduction Large language models (LLMs), such as OpenAI's ChatGPT and similar chatbot products from other organizations, have recently gained widespread adoption. These models can extend text or respond to instructions in a natural and helpful manner. Despite the core technologies behind LLMs, nam...

    A Call to Protect Open-Source AI in Europe

    by: LAION.ai, 28 Apr, 2023

    An Open Letter to the European Parliament: Protecting Open-Source AI for a Safe, Secure, and Sovereign Digital Future LAION, alongside prominent research institutions and developers, has penned an open letter to the European Parliament to express concerns about the draft AI Act's potential impact on...

    Training a Binary Classifier to Distinguish Images Generated with Stable Diffusion (v1.4) from Real Ones

    by: Christoph Schuhmann, Ilia Zaitsev, 12 Apr, 2023

    We present the development and assessment of a binary classifier designed to distinguish between authentic images and images generated using Stable Diffusion (SD) v1.4. We will discuss the dataset employed, describe the model architecture, outline the training process, and present the results obtain...

    General-GPT: Breaking the Modality Constraint

    by: Shivaen Ramshetty and Christoph Schuhmann, 28 Mar, 2023

    Introduction -With the rapid explosion of large language models and utilization of their encompassing applications, most notably ChatGPT, there is a clear promise of more capable and useful AI models/systems. Often, such models are compared to us as humans using the Turing test or their performance o...

    \ No newline at end of file +With the rapid explosion of large language models and utilization of their encompassing applications, most notably ChatGPT, there is a clear promise of more capable and useful AI models/systems. Often, such models are compared to us as humans using the Turing test or their performance o...

    \ No newline at end of file diff --git a/notes/laion-debate/index.html b/notes/laion-debate/index.html index fa821520..c57b7f10 100644 --- a/notes/laion-debate/index.html +++ b/notes/laion-debate/index.html @@ -1,4 +1,4 @@ -Call to Build Open Multi-Modal Models for Personal Assistants | LAION

    LAION

    CALL TO BUILD OPEN MULTI-MODAL MODELS FOR PERSONAL ASSISTANTS

    by: LAION, 28 Jun, 2024

    We’re pleased to announce the World's first Large Competitive Debate Dataset: LAION-Debate. LAION-Debate is a large Competitive debate dataset providing links to Competitive Debate Championships, discussions and prominent speakers intake and conversations posted on YouTube by University of Cambridge and University of Oxford through their Cambridge and Oxford Union Debate clubs on their affiliated channels.

    +Call to Build Open Multi-Modal Models for Personal Assistants | LAION

    CALL TO BUILD OPEN MULTI-MODAL MODELS FOR PERSONAL ASSISTANTS

    by: LAION, 28 Jun, 2024

    We’re pleased to announce the World's first Large Competitive Debate Dataset: LAION-Debate. LAION-Debate is a large Competitive debate dataset providing links to Competitive Debate Championships, discussions and prominent speakers intake and conversations posted on YouTube by University of Cambridge and University of Oxford through their Cambridge and Oxford Union Debate clubs on their affiliated channels.

    Competitive Debate datasets are scarce and hard to find in the public domain. Because these datasets are either gated by individuals and institutions who generate them or not archived properly enough to form them into a dataset. Hindering the ability to use them for Artificial Intelligence research.

    In an era, where datasets are being scarce and the large AI models are exhausting entire human knowledge and depleting known data sources, Debate 2B encourages to use alternative credible sources and other forms of knowledge corpus that provides a unique outlook and understanding than the mainstream.

    Today, a community member of LAION (tawsif) released this novel dataset on Competitive Debate in the field of Natural Language Processing.

    @@ -30,4 +30,4 @@

    https://github.com/sleepingcat4 Email: tawsif.ahmed@science.ru.nl

    -

    \ No newline at end of file +
    \ No newline at end of file diff --git a/notes/laion-maintenance/index.html b/notes/laion-maintenance/index.html index d391a392..fdb517cb 100644 --- a/notes/laion-maintenance/index.html +++ b/notes/laion-maintenance/index.html @@ -1,7 +1,7 @@ -Safety Review for LAION 5B | LAION

    LAION

    SAFETY REVIEW FOR LAION 5B

    by: LAION.ai, 19 Dec, 2023

    There have been reports in the press about the results of a research project at Stanford University, according to which the LAION training set 5B contains potentially illegal content in the form of CSAM. We would like to comment on this as follows:

    +Safety Review for LAION 5B | LAION

    SAFETY REVIEW FOR LAION 5B

    by: LAION.ai, 19 Dec, 2023

    There have been reports in the press about the results of a research project at Stanford University, according to which the LAION training set 5B contains potentially illegal content in the form of CSAM. We would like to comment on this as follows:

    LAION is a non-profit organization that provides datasets, tools and models for the advancement of machine learning research. We are committed to open public education and the environmentally safe use of resources through the reuse of existing datasets and models.

    LAION datasets (more than 5.85 billion entries) are sourced from the freely available Common Crawl web index and offer only links to content on the public web, with no images. We developed and published our own rigorous filters to detect and remove illegal content from LAION datasets before releasing them.

    LAION collaborates with universities, researchers and NGOs to improve these filters and are currently working with the Internet Watch Foundation (IWF) to identify and remove content suspected of violating laws. LAION invites the Stanford researchers to join its Community to improve our datasets and to develop efficient filters for detecting harmful content.

    LAION has a zero tolerance policy for illegal content and in an abundance of caution, we are temporarily taking down the LAION datasets to ensure they are safe before republishing them.

    Following a discussion with the Hamburg State Data Protection Commissioner, we would also like to point out that the CSAM data is data that must be deleted immediately for data protection reasons in accordance with Art. 17 GDPR.

    -
    \ No newline at end of file +
    \ No newline at end of file diff --git a/notes/letter-to-the-eu-parliament/index.html b/notes/letter-to-the-eu-parliament/index.html index b05433d6..c6f881b2 100644 --- a/notes/letter-to-the-eu-parliament/index.html +++ b/notes/letter-to-the-eu-parliament/index.html @@ -1,7 +1,7 @@ A Call to Protect Open-Source AI in Europe | LAION

    LAION

    A CALL TO PROTECT OPEN-SOURCE AI IN EUROPE

    by: LAION.ai, 28 Apr, 2023

    An Open Letter to the European Parliament: Protecting Open-Source AI for a Safe, Secure, and Sovereign Digital Future

    +<p>LAION, along..."/>

    A CALL TO PROTECT OPEN-SOURCE AI IN EUROPE

    by: LAION.ai, 28 Apr, 2023

    An Open Letter to the European Parliament: Protecting Open-Source AI for a Safe, Secure, and Sovereign Digital Future

    LAION, alongside prominent research institutions and developers, has penned an open letter to the European Parliament to express concerns about the draft AI Act's potential impact on open-source research and development (R&D) in artificial intelligence (AI). The letter highlights the importance of open-source R&D for ensuring the safety, security, and competitiveness of AI in Europe and warns against the consequences of stifling such innovation.

    The Importance of Open-Source AI

    The letter outlines three main reasons why open-source AI is worth protecting:

    @@ -59,4 +59,4 @@

    CALL TO BUILD OPEN MULTI-MODAL MODELS FOR PERSONAL ASSISTANTS

    by: Christoph Schuhmann, 29 May, 2024

    Technologies like the recently introduced GPT-4-OMNI from OpenAI show again the potential which strong multi-modal models might have to positively transform many aspects of our lives. A particularly impressive example of this is in the field of education. Imagine every person in the world having their own personal learning assistant that acts like a attentive, caring, patient, and empathetic tutor. The demo from OpenAI last Monday showed that such a vision of the future is not too far off and is within reach.

    +Call to Build Open Multi-Modal Models for Personal Assistants | LAION

    CALL TO BUILD OPEN MULTI-MODAL MODELS FOR PERSONAL ASSISTANTS

    by: Christoph Schuhmann, 29 May, 2024

    Technologies like the recently introduced GPT-4-OMNI from OpenAI show again the potential which strong multi-modal models might have to positively transform many aspects of our lives. A particularly impressive example of this is in the field of education. Imagine every person in the world having their own personal learning assistant that acts like a attentive, caring, patient, and empathetic tutor. The demo from OpenAI last Monday showed that such a vision of the future is not too far off and is within reach.

    The Path to Open Multi-Modal Models

    An important milestone on this path could be training an open-source model with capabilities similar to GPT-4-OMNI. The first step would be to fine-tune an existing large language model so that it can natively understand and process audio in the same way large language models currently handle text. Simultaneously, this model should be able to generate audio natively, just as it can currently output and manipulate text.

    This approach had been shown to work in the AudioPalm paper:

    @@ -46,4 +46,4 @@

    Join our discord server

    -

    \ No newline at end of file +
    \ No newline at end of file diff --git a/notes/realfake/index.html b/notes/realfake/index.html index 43da64eb..642b46eb 100644 --- a/notes/realfake/index.html +++ b/notes/realfake/index.html @@ -1,7 +1,7 @@ Training a Binary Classifier to Distinguish Images Generated with Stable Diffusion (v1.4) from Real Ones | LAION

    LAION

    TRAINING A BINARY CLASSIFIER TO DISTINGUISH IMAGES GENERATED WITH STABLE DIFFUSION (V1.4) FROM REAL ONES

    by: Christoph Schuhmann, Ilia Zaitsev, 12 Apr, 2023

    We present the development and assessment of a binary classifier designed to distinguish between authentic images and images generated +using Stable Diffus..."/>

    TRAINING A BINARY CLASSIFIER TO DISTINGUISH IMAGES GENERATED WITH STABLE DIFFUSION (V1.4) FROM REAL ONES

    by: Christoph Schuhmann, Ilia Zaitsev, 12 Apr, 2023

    We present the development and assessment of a binary classifier designed to distinguish between authentic images and images generated using Stable Diffusion (SD) v1.4. We will discuss the dataset employed, describe the model architecture, outline the training process, and present the results obtained. Furthermore, we will explore potential future work aimed at enhancing the classifier's performance. The source code, training parameters, and model weights are available in this repository.

    @@ -83,4 +83,4 @@

    PyTorch-Lightning.
  • Numerous other open-source tools, models, and datasets made this work possible.
    • -

    \ No newline at end of file +
    \ No newline at end of file diff --git a/press/index.html b/press/index.html index 1fba299b..72344006 100644 --- a/press/index.html +++ b/press/index.html @@ -1 +1 @@ -Press | LAION

    LAION

    PRESS RELEASES

    30 August, 2024 - ARSTECHNICA.COM

    https://www.arstechnica.com/tech-policy/2024/08/nonprofit-scrubs-illegal-content-from-controversial-ai-training-dataset

    10 June, 2024 - HEISE.DE

    https://www.heise.de/news/Reasoning-Fail-Gaengige-LLMs-scheitern-an-kinderleichter-Aufgabe-9755034.html

    16 February, 2024 - NATURE.COM

    https://www.nature.com/articles/d41586-024-00497-8

    15 February, 2024 - TECHCRUNCH.COM

    https://www.techcrunch.com/2024/02/15/this-german-nonprofit-is-building-an-open-voice-assistant-that-anyone-can-use/

    14 February, 2024 - HEISE.DE

    https://www.heise.de/news/KI-Update-kompakt-ChatGPT-personalisieren-Digitalisierung-IAFM-Bud-E-9628384.html

    13 February, 2024 - AIBUSINESS.COM

    https://www.aibusiness.com/nlp/bud-e-aims-to-create-human-like-computer-voices

    13 February, 2024 - HEISE.DE

    https://www.heise.de/news/Bud-E-Laion-arbeitet-an-Audio-KI-mit-Empathie-9626286.html

    30 November, 2023 - FORBES.COM

    https://www.forbes.com/sites/hessiejones/2023/11/30/the-rise-and-fall-and-rise-of-sam-altman-has-grave-implications-for-ai-research-and-humanity/

    24 November, 2023 - SPEKTRUM.DE

    https://www.spektrum.de/news/interview-mit-laion-ki-im-sinn-der-menschheit-erforschen/2198369

    27 October, 2023 - TECHCRUNCH.COM

    https://www.techcrunch.com/2023/10/27/a-group-behind-stable-diffusion-wants-to-open-source-emotion-detecting-ai

    29 September, 2023 - GOLEM.DE

    https://www.golem.de/news/laion-riesiges-sprachmodell-fuer-deutsch-trainiert-2309-178081.html

    31 August, 2023 - APNEWS.COM

    https://www.apnews.com/article/artists-ai-image-generators-stable-diffusion-midjourney-7ebcb6e6ddca3f165a3065c70ce85904

    16 July, 2023 - ARDMEDIATHEK.DE

    https://www.ardmediathek.de/video/ttt-titel-thesen-temperamente/verliert-europa-den-anschluss/das-erste/Y3JpZDovL2Rhc2Vyc3RlLmRlL3R0dCAtIHRpdGVsIHRoZXNlbiB0ZW1wZXJhbWVudGUvYjMxZTE1NDAtM2M4OC00MGExLWJiNjQtODhlYTgzNzg0MzI5

    16 July, 2023 - REPUBBLICA.IT

    https://www.repubblica.it/economia/2023/07/17/news/maghi_tedeschi_algoritmi_ai-407769374/

    7 July, 2023 - TAGESSCHAU.DE

    https://www.tagesschau.de/wissen/technologie/ki-trainingsdaten-privat-datenschutz-100.html

    15 June, 2023 - GOLEM.DE

    https://www.golem.de/news/interview-mit-laion-der-datensatz-hinter-den-ki-bildern-2306-174786.html

    4 May, 2023 - THEGUARDIAN.COM

    https://www.theguardian.com/technology/2023/may/04/eu-urged-to-protect-grassroots-ai-research-or-risk-losing-out-to-us

    28 April, 2023 - UNITE.AI

    https://www.unite.ai/laion-and-a-group-of-27/

    26 April, 2023 - SIFTED.EU

    https://www.sifted.eu/articles/ai-supercomputer-petition-stable-diffusion

    24 April, 2023 - BLOOMBERG.COM

    https://www.bloomberg.com/news/features/2023-04-24/a-high-school-teacher-s-free-image-database-powers-ai-unicorns#xj4y7vzkg

    19 April, 2023 - FORBES.COM

    https://www.forbes.com/sites/hessiejones/2023/04/19/amid-growing-call-to-pause-ai-research-laion-petitions-governments-to-keep-agi-research-open-active-and-responsible/

    17 April, 2023 - SUEDDEUTSCHE.DE

    https://www.sueddeutsche.de/wirtschaft/ki-laion-kuenstliche-intelligenz-supercomputer-open-ai-chatgpt-1.5804840?reduced=true

    8 April, 2023 - HEISE.DE

    https://www.heise.de/news/Open-Source-KI-LAION-ruft-zu-offenem-GPT-4-Nachbau-auf-genug-Wissen-sei-da-8651446.html

    4 March, 2023 - SPIEGEL.DE

    https://www.spiegel.de/spiegel/print/index-2023-10.html

    13 December, 2022 - GOLEM.DE

    https://www.golem.de/news/laion-5b-datensatz-zum-ki-training-bietet-fast-6-milliarden-bilder-2212-170470.html
    \ No newline at end of file +Press | LAION

    LAION

    PRESS RELEASES

    30 August, 2024 - ARSTECHNICA.COM

    https://www.arstechnica.com/tech-policy/2024/08/nonprofit-scrubs-illegal-content-from-controversial-ai-training-dataset

    30 August, 2024 - HEISE.DE

    https://www.heise.de/en/news/RE-LAION-5B-Image-database-without-abuse-images-9852925.html

    30 August, 2024 - TECHCRUNCH.COM

    https://www.techcrunch.com/2024/08/30/the-org-behind-the-data-set-used-to-train-stable-diffusion-claims-it-has-removed-csam/

    30 August, 2024 - APNEWS.COM

    https://www.apnews.com/article/ai-image-generators-child-sexual-abuse-laion-stable-diffusion-2652b0f4245fb28ced1cf74c60a8d9f0

    10 June, 2024 - HEISE.DE

    https://www.heise.de/news/Reasoning-Fail-Gaengige-LLMs-scheitern-an-kinderleichter-Aufgabe-9755034.html

    16 February, 2024 - NATURE.COM

    https://www.nature.com/articles/d41586-024-00497-8

    15 February, 2024 - TECHCRUNCH.COM

    https://www.techcrunch.com/2024/02/15/this-german-nonprofit-is-building-an-open-voice-assistant-that-anyone-can-use/

    14 February, 2024 - HEISE.DE

    https://www.heise.de/news/KI-Update-kompakt-ChatGPT-personalisieren-Digitalisierung-IAFM-Bud-E-9628384.html

    13 February, 2024 - AIBUSINESS.COM

    https://www.aibusiness.com/nlp/bud-e-aims-to-create-human-like-computer-voices

    13 February, 2024 - HEISE.DE

    https://www.heise.de/news/Bud-E-Laion-arbeitet-an-Audio-KI-mit-Empathie-9626286.html

    30 November, 2023 - FORBES.COM

    https://www.forbes.com/sites/hessiejones/2023/11/30/the-rise-and-fall-and-rise-of-sam-altman-has-grave-implications-for-ai-research-and-humanity/

    24 November, 2023 - SPEKTRUM.DE

    https://www.spektrum.de/news/interview-mit-laion-ki-im-sinn-der-menschheit-erforschen/2198369

    27 October, 2023 - TECHCRUNCH.COM

    https://www.techcrunch.com/2023/10/27/a-group-behind-stable-diffusion-wants-to-open-source-emotion-detecting-ai

    29 September, 2023 - GOLEM.DE

    https://www.golem.de/news/laion-riesiges-sprachmodell-fuer-deutsch-trainiert-2309-178081.html

    31 August, 2023 - APNEWS.COM

    https://www.apnews.com/article/artists-ai-image-generators-stable-diffusion-midjourney-7ebcb6e6ddca3f165a3065c70ce85904

    16 July, 2023 - ARDMEDIATHEK.DE

    https://www.ardmediathek.de/video/ttt-titel-thesen-temperamente/verliert-europa-den-anschluss/das-erste/Y3JpZDovL2Rhc2Vyc3RlLmRlL3R0dCAtIHRpdGVsIHRoZXNlbiB0ZW1wZXJhbWVudGUvYjMxZTE1NDAtM2M4OC00MGExLWJiNjQtODhlYTgzNzg0MzI5

    16 July, 2023 - REPUBBLICA.IT

    https://www.repubblica.it/economia/2023/07/17/news/maghi_tedeschi_algoritmi_ai-407769374/

    7 July, 2023 - TAGESSCHAU.DE

    https://www.tagesschau.de/wissen/technologie/ki-trainingsdaten-privat-datenschutz-100.html

    15 June, 2023 - GOLEM.DE

    https://www.golem.de/news/interview-mit-laion-der-datensatz-hinter-den-ki-bildern-2306-174786.html

    4 May, 2023 - THEGUARDIAN.COM

    https://www.theguardian.com/technology/2023/may/04/eu-urged-to-protect-grassroots-ai-research-or-risk-losing-out-to-us

    28 April, 2023 - UNITE.AI

    https://www.unite.ai/laion-and-a-group-of-27/

    26 April, 2023 - SIFTED.EU

    https://www.sifted.eu/articles/ai-supercomputer-petition-stable-diffusion

    24 April, 2023 - BLOOMBERG.COM

    https://www.bloomberg.com/news/features/2023-04-24/a-high-school-teacher-s-free-image-database-powers-ai-unicorns#xj4y7vzkg

    19 April, 2023 - FORBES.COM

    https://www.forbes.com/sites/hessiejones/2023/04/19/amid-growing-call-to-pause-ai-research-laion-petitions-governments-to-keep-agi-research-open-active-and-responsible/

    17 April, 2023 - SUEDDEUTSCHE.DE

    https://www.sueddeutsche.de/wirtschaft/ki-laion-kuenstliche-intelligenz-supercomputer-open-ai-chatgpt-1.5804840?reduced=true

    8 April, 2023 - HEISE.DE

    https://www.heise.de/news/Open-Source-KI-LAION-ruft-zu-offenem-GPT-4-Nachbau-auf-genug-Wissen-sei-da-8651446.html

    4 March, 2023 - SPIEGEL.DE

    https://www.spiegel.de/spiegel/print/index-2023-10.html

    13 December, 2022 - GOLEM.DE

    https://www.golem.de/news/laion-5b-datensatz-zum-ki-training-bietet-fast-6-milliarden-bilder-2212-170470.html
    \ No newline at end of file diff --git a/privacy-policy/index.html b/privacy-policy/index.html index 4ef9bf64..c979c9de 100644 --- a/privacy-policy/index.html +++ b/privacy-policy/index.html @@ -1 +1 @@ -Privacy Policy | LAION

    LAION

    PRIVACY POLICY

    Overview

    We are pleased that you are interested in our work and welcome you to our website laion.ai. In this Privacy Policy you will learn which personal data we process when you visit our website and to what kind of purpose, and also what rights you have regarding these data. Categorically, we only store data as long as we need them. There is no legal obligation to provide us with personal data. Automated decision-making, as per Article 22 of the EU-GDPR, will not happen.

    1. Definitions

    We are required by law that personal data are processed lawfully, in good faith, and in a manner that can be comprehended by the persons who are affected (“lawfulness, fair processing, transparency”). To this end, we hereby inform you about the individual legal definitions of the European General Data Protection Regulation (GDPR) and the new German Federal Data Protection Act, which are also used in these data privacy regulations.

    1.1 Personal data

    'Personal data' means any information relating to an identified or identifiable natural person (hereinafter the 'data subject'). A natural person is considered to be identifiable if he or she can be identified directly or indirectly, in particular by association with an identifier such as a name, an identification number, location data, an online identifier, or one or more special features which express the physical, physiological, genetic, mental, economic, cultural or social identity of the natural person.

    1.3 Restriction of processing

    'Restriction of processing' means the marking of stored personal data with the aim of limiting its processing in the future.

    1.4 Profiling

    'Profiling' means any form of automated processing of personal data consisting of the use of personal data to evaluate certain personal aspects relating to a natural person, in particular to analyse or predict aspects concerning that natural person's performance at work, economic situation, health, personal preferences, interests, reliability, behaviour, location or movements.

    1.5 Pseudonymization

    'Pseudonymization' means the processing of personal data in such a manner that the personal data can no longer be attributed to a specific data subject without the use of additional information, provided that such additional information is kept separately and is subject to technical and organizational measures to ensure that the personal data is not attributed to an identified or identifiable natural person

    1.6 Filing system

    'Filing system' means any structured set of personal data which is accessible according to specific criteria, whether centralized, decentralized or dispersed on a functional or geographical basis.

    1.7 Controller

    'Controller' means the natural or legal person, public authority, agency or other body which, alone or jointly with others, determines the purposes and means of the processing of personal data. Where the purposes and means of such processing are determined by European Union or Member State law, the controller or the specific criteria for its nomination may be provided for by European Union or Member State law.

    1.8 Processor

    'Processor' means a natural or legal person, public authority, agency or other body which processes personal data on behalf of the controller.

    1.9 Recipient

    'Recipient' means a natural or legal person, public authority, agency or another body, to which the personal data is disclosed, whether a third party or not. However, public authorities which may receive potentially personal data in the framework of a particular inquiry in accordance with European Union or Member State law shall not be regarded as recipients. The processing of that data by those public authorities shall be in compliance with the applicable data protection rules according to the purposes of the processing.

    1.10 Third party

    A 'third party' means a natural or legal person, public authority, agency or body other than the data subject, controller, processor and persons who, under the direct authority of the controller or processor, are authorized to process personal data.

    2. Responsible controller

    Responsible controller is: LAION e.V., Marie-Henning-Weg 143, 21035 Hamburg, Germany

    3. Data we collect

    During the use of our website, we do not collect any data except when the user fills out forms.

    4. Inquiries

    When you contact us via e-mail, telephone or telefax, your inquiry, including all personal data arising thereof will be stored by us for the purpose of processing your request. We will not pass on these data without your consent. The processing of these data is based on Article 6 (1) (1) (b) GDPR, if your inquiry is related to the fulfilment of a contract concluded with us or required for the implementation of pre-contractual measures. Furthermore, the processing is based on Article 6 (1) (1) (f) GDPR, because we have a legitimate interest in the effective handling of requests sent to us. In addition, according to Article 6 (1) (1) (c) GDPR we are also entitled to the processing of the above-mentioned data, because we are legally bound to enable fast electronic contact and immediate communication. Of course, your data will only be used strictly according to purpose and only for processing and responding to your request. After final processing, your data will immediately be anonymized or deleted, unless we are bound by a legally prescribed storage period.

    5. Processors

    In principle, we will never pass on your personal data to third parties without your explicit consent. However, just as every modern business we cooperate with data processors in order to be able to offer you the best possible uninterrupted service. When we cooperate with external service providers, regular order processing is performed, based on Article 28 GDPR. For this purpose, we enter into respective agreements with our partners, in order to safeguard the protection of your data. For processing your data, we only use carefully selected processors. They are bound by our instructions, and regularly controlled by us. We only commission external service provider who have guaranteed that all data processing procedures are performed in unison with data protection regulations. Receivers of personal data may be: Hosting companies and Hosting service providers

    6. Children and young people

    In principle, our offer is directed towards adults. Children and young people under the age of 16 are not allowed to transmit personal data to us without the consent of their parents or legal guardians.

    7. Your rights

    If your personal data is processed on the basis of consent which you have given us, you have the right to revoke your consent at any time. The revocation of consent does not affect the legality of the processing performed on the basis of the consent until the time of revocation. You can contact us at any time to exercise your right to revoke consent.

    7.2 Right to confirmation

    You have the right to request confirmation from the controller that we are processing personal data concerning you. You can request this confirmation at any time using the contact details above.

    7.3 Right to information

    In the event that personal data is processed, you can request information about this personal data and the following information at any time: the purposes of the processing, the categories of personal data being processed, the recipients or categories of recipients to whom the personal data has been or is being disclosed, in particular in the case of recipients in third countries or international organizations, if possible, the planned duration for which the personal data is stored or, if this is not, possible, the criteria for determining this duration, the existence of a right to rectification or erasure of the personal data concerning you, or to a restriction of processing by the controller or a right to object to such processing, the existence of a right to lodge a complaint with a supervisory authority, if the personal data is not collected from the data subject, all available information on the source of the data, the existence of automated decision-making, including profiling, in accordance with Article 22 (1) and (4) GDPR and, at least in these cases, meaningful information about the logic involved and the scope and intended impact of such processing on the data subject. If personal data is transferred to a third country or to an international organization, you have the right to be informed of the appropriate safeguards under Article 46 of the GDPR in connection with the transfer. We provide a copy of the personal data that is the subject of the processing. For any additional copies you request of a person, we may charge a reasonable fee based on our administrative costs. If your request is submitted electronically, the information must be provided in a standard electronic format, unless otherwise stated. The right to receive a copy under paragraph 3 shall not affect the rights and freedoms of others.

    7.4 Right to rectification

    You have the right to demand the immediate correction of incorrect personal data concerning you. Taking into account the purposes of processing, you have the right to request the completion of incomplete personal data, including by means of a supplementary statement.

    7.4 Right to rectification

    You have the right to demand the immediate correction of incorrect personal data concerning you. Taking into account the purposes of processing, you have the right to request the completion of incomplete personal data, including by means of a supplementary statement.

    7.5 Right to erasure (“right to be forgotten“)

    You have the right to demand that the controller erase personal data concerning you without undue delay, and we are obligated to erase personal data without undue delay where one of the following grounds applies: the personal data are no longer necessary in relation to the purposes for which they were collected or otherwise processed, the data subject withdraws the consent on which the processing is based according to point (a) of Article 6(1), or point (a) of Article 9(2), and there is no other legal ground for the processing, the data subject objects to the processing pursuant to Article 21(1) GDPR and there are no overriding legitimate grounds for the processing, or the data subject objects to the processing pursuant to Article 21(2) GDPR, the personal data have been unlawfully processed, personal data must be erased for compliance with a legal obligation in Union or Member State law to which the controller is subject, the personal data was collected in relation to the offer of information society services referred to in Article 8(1) GDPR. If the controller has made the personal data public and is obliged pursuant to paragraph 1 to erase the personal data, the controller, taking account of available technology and the cost of implementation, shall take reasonable steps, including technical measures, to inform controllers which are processing the personal data that the data subject has requested the erasure by such controllers of any links to, or copy or replication of, that personal data. The right to erasure (“right to be forgotten“) does not apply to the extent that the processing is necessary: to exercise the right of freedom of expression and information, for compliance with a legal obligation which requires processing by Union or Member, State law to which the controller is subject or for the performance of a task carried out in the public interest or in the exercise of official authority vested in the controller, for reasons of public interest in the area of public health in accordance with points (h) and (i) of Article 9(2) as well as Article 9(3) GDPR, for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes in accordance with Article 89(1) GDPR in so far as the right referred to in paragraph 1 is likely to render impossible or seriously impair the achievement of the objectives of that processing; or for the establishment, exercise or defense of legal claims

    7.6 Right to restriction of processing

    You have the right to request that we restrict the processing of your personal data if any of the following conditions apply: the accuracy of the personal data is contested by the data subject, for a period enabling the controller to verify the accuracy of the personal data, the processing is unlawful and the data subject opposes the erasure of the personal data and requests the restriction of their use instead, the controller no longer needs the personal data for the purposes of the processing, but the data is required by the data subject for the establishment, exercise or defense of legal claims, or the data subject has objected to processing pursuant to Article 21(1) GDPR pending the verification whether the legitimate grounds of the controller override those of the data subject In the event that processing has been restricted under the aforementioned conditions, this personal data shall – with the exception of storage – only be processed with the data subject’s consent or for the establishment, exercise or defense of legal claims or for the protection of the rights of another natural or legal person or for reasons of important public interest of the Union or of a Member State. In order to exercise the right to restrict processing, the data subject may contact us at any time using the contact details provided above.

    7.7 Right to data portability

    You have the right to receive the personal data concerning you which you have provided to us in a structured, commonly used and machine-readable format and have the right to transmit that data to another controller without hindrance from the controller to which the personal data have been provided, to the extent that: the processing is based on consent pursuant to point (a) of Article 6 (1) or point (a) of Article 9 (2) or on a contract pursuant to point (b) of Article 6 (1) GDPR and the processing is carried out by automated means. In exercising your right to data portability pursuant to paragraph 1, you have the right to have the personal data transmitted directly from one controller to another, to the extent that this is technically feasible. The exercise of the right to data portability does not affect your right to erasure (“right to be forgotten”). That right shall not apply to processing necessary for the performance of a task carried out in the public interest or in the exercise of official authority vested in the controller.

    7.8 Right to object

    You have the right to object, on grounds relating to your particular situation, at any time to processing of personal data which concerns you which is based on point (e) or (f) of Article 6 (1) GDPR, including profiling based on those provisions. If objection is made, the controller will no longer process the personal data unless the controller demonstrates compelling legitimate grounds for the processing which override the interests, rights and freedoms of the data subject or for the establishment, exercise or defense of legal claims. In the event that personal data is processed for direct marketing purposes, you have the right to object at any time to processing of personal data concerning you for such marketing. This also applies to profiling to the extent that it is related to such direct marketing. If you object to processing for direct marketing purposes, your personal data shall no longer be processed for such purposes. Regarding the use of information society services, and notwithstanding Directive 2002/58/EC, you can exercise your right to object by automated means using technical specifications. Where personal data are processed for scientific or historical research purposes or statistical purposes pursuant to Article 89 (1), you, on grounds relating to your particular situation, have the right to object to processing of personal data concerning you, unless the processing is necessary for the performance of a task carried out for reasons of public interest. The right of objection can be exercised at any time by contacting the respective controller.

    7.9 Automated individual decision-making, including profiling

    You have the right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects for you or similarly significantly affects you. This does not apply if the decision: is necessary for entering into, or performance of, a contract between the data subject and a data controller, is authorized by Union or Member State law to which the controller is subject and which also lays down suitable measures to safeguard the data subject’s rights and freedoms and legitimate interests, or is based on the data subject’s explicit consent. The controller shall implement suitable measures to safeguard the data subject’s rights and freedoms and legitimate interests, at least the right to obtain human intervention on the part of the controller, to express his or her point of view and to contest the decision. This right can be exercised by the data subject at any time by contacting the respective controller.

    7.10 Right to lodge a complaint with a supervisory authority

    You also have the right, without prejudice to any other administrative or judicial remedy, to lodge a complaint with a supervisory authority, in particular in the Member State of your habitual residence, place of work or place of the alleged infringement if you as data subject consider that the processing of personal data relating to you infringes this Regulation.

    7.11 Right to effective judicial remedy

    Without prejudice to any other available administrative or judicial remedy, including the right to lodge a complaint with a supervisory authority pursuant to Article 77 GDPR, you have the right to an effective judicial remedy if you consider that your rights under this Regulation have been infringed as a result of the processing of your personal data in breach of this Regulation.

    Submitting requests

    If you have any privacy concerns related to this website and the provided datasets or want to have taken images out of the dataset, fill out the following form.

    \ No newline at end of file +Privacy Policy | LAION

    LAION

    PRIVACY POLICY

    Overview

    We are pleased that you are interested in our work and welcome you to our website laion.ai. In this Privacy Policy you will learn which personal data we process when you visit our website and to what kind of purpose, and also what rights you have regarding these data. Categorically, we only store data as long as we need them. There is no legal obligation to provide us with personal data. Automated decision-making, as per Article 22 of the EU-GDPR, will not happen.

    1. Definitions

    We are required by law that personal data are processed lawfully, in good faith, and in a manner that can be comprehended by the persons who are affected (“lawfulness, fair processing, transparency”). To this end, we hereby inform you about the individual legal definitions of the European General Data Protection Regulation (GDPR) and the new German Federal Data Protection Act, which are also used in these data privacy regulations.

    1.1 Personal data

    'Personal data' means any information relating to an identified or identifiable natural person (hereinafter the 'data subject'). A natural person is considered to be identifiable if he or she can be identified directly or indirectly, in particular by association with an identifier such as a name, an identification number, location data, an online identifier, or one or more special features which express the physical, physiological, genetic, mental, economic, cultural or social identity of the natural person.

    1.3 Restriction of processing

    'Restriction of processing' means the marking of stored personal data with the aim of limiting its processing in the future.

    1.4 Profiling

    'Profiling' means any form of automated processing of personal data consisting of the use of personal data to evaluate certain personal aspects relating to a natural person, in particular to analyse or predict aspects concerning that natural person's performance at work, economic situation, health, personal preferences, interests, reliability, behaviour, location or movements.

    1.5 Pseudonymization

    'Pseudonymization' means the processing of personal data in such a manner that the personal data can no longer be attributed to a specific data subject without the use of additional information, provided that such additional information is kept separately and is subject to technical and organizational measures to ensure that the personal data is not attributed to an identified or identifiable natural person

    1.6 Filing system

    'Filing system' means any structured set of personal data which is accessible according to specific criteria, whether centralized, decentralized or dispersed on a functional or geographical basis.

    1.7 Controller

    'Controller' means the natural or legal person, public authority, agency or other body which, alone or jointly with others, determines the purposes and means of the processing of personal data. Where the purposes and means of such processing are determined by European Union or Member State law, the controller or the specific criteria for its nomination may be provided for by European Union or Member State law.

    1.8 Processor

    'Processor' means a natural or legal person, public authority, agency or other body which processes personal data on behalf of the controller.

    1.9 Recipient

    'Recipient' means a natural or legal person, public authority, agency or another body, to which the personal data is disclosed, whether a third party or not. However, public authorities which may receive potentially personal data in the framework of a particular inquiry in accordance with European Union or Member State law shall not be regarded as recipients. The processing of that data by those public authorities shall be in compliance with the applicable data protection rules according to the purposes of the processing.

    1.10 Third party

    A 'third party' means a natural or legal person, public authority, agency or body other than the data subject, controller, processor and persons who, under the direct authority of the controller or processor, are authorized to process personal data.

    2. Responsible controller

    Responsible controller is: LAION e.V., Marie-Henning-Weg 143, 21035 Hamburg, Germany

    3. Data we collect

    During the use of our website, we do not collect any data except when the user fills out forms.

    4. Inquiries

    When you contact us via e-mail, telephone or telefax, your inquiry, including all personal data arising thereof will be stored by us for the purpose of processing your request. We will not pass on these data without your consent. The processing of these data is based on Article 6 (1) (1) (b) GDPR, if your inquiry is related to the fulfilment of a contract concluded with us or required for the implementation of pre-contractual measures. Furthermore, the processing is based on Article 6 (1) (1) (f) GDPR, because we have a legitimate interest in the effective handling of requests sent to us. In addition, according to Article 6 (1) (1) (c) GDPR we are also entitled to the processing of the above-mentioned data, because we are legally bound to enable fast electronic contact and immediate communication. Of course, your data will only be used strictly according to purpose and only for processing and responding to your request. After final processing, your data will immediately be anonymized or deleted, unless we are bound by a legally prescribed storage period.

    5. Processors

    In principle, we will never pass on your personal data to third parties without your explicit consent. However, just as every modern business we cooperate with data processors in order to be able to offer you the best possible uninterrupted service. When we cooperate with external service providers, regular order processing is performed, based on Article 28 GDPR. For this purpose, we enter into respective agreements with our partners, in order to safeguard the protection of your data. For processing your data, we only use carefully selected processors. They are bound by our instructions, and regularly controlled by us. We only commission external service provider who have guaranteed that all data processing procedures are performed in unison with data protection regulations. Receivers of personal data may be: Hosting companies and Hosting service providers

    6. Children and young people

    In principle, our offer is directed towards adults. Children and young people under the age of 16 are not allowed to transmit personal data to us without the consent of their parents or legal guardians.

    7. Your rights

    If your personal data is processed on the basis of consent which you have given us, you have the right to revoke your consent at any time. The revocation of consent does not affect the legality of the processing performed on the basis of the consent until the time of revocation. You can contact us at any time to exercise your right to revoke consent.

    7.2 Right to confirmation

    You have the right to request confirmation from the controller that we are processing personal data concerning you. You can request this confirmation at any time using the contact details above.

    7.3 Right to information

    In the event that personal data is processed, you can request information about this personal data and the following information at any time: the purposes of the processing, the categories of personal data being processed, the recipients or categories of recipients to whom the personal data has been or is being disclosed, in particular in the case of recipients in third countries or international organizations, if possible, the planned duration for which the personal data is stored or, if this is not, possible, the criteria for determining this duration, the existence of a right to rectification or erasure of the personal data concerning you, or to a restriction of processing by the controller or a right to object to such processing, the existence of a right to lodge a complaint with a supervisory authority, if the personal data is not collected from the data subject, all available information on the source of the data, the existence of automated decision-making, including profiling, in accordance with Article 22 (1) and (4) GDPR and, at least in these cases, meaningful information about the logic involved and the scope and intended impact of such processing on the data subject. If personal data is transferred to a third country or to an international organization, you have the right to be informed of the appropriate safeguards under Article 46 of the GDPR in connection with the transfer. We provide a copy of the personal data that is the subject of the processing. For any additional copies you request of a person, we may charge a reasonable fee based on our administrative costs. If your request is submitted electronically, the information must be provided in a standard electronic format, unless otherwise stated. The right to receive a copy under paragraph 3 shall not affect the rights and freedoms of others.

    7.4 Right to rectification

    You have the right to demand the immediate correction of incorrect personal data concerning you. Taking into account the purposes of processing, you have the right to request the completion of incomplete personal data, including by means of a supplementary statement.

    7.4 Right to rectification

    You have the right to demand the immediate correction of incorrect personal data concerning you. Taking into account the purposes of processing, you have the right to request the completion of incomplete personal data, including by means of a supplementary statement.

    7.5 Right to erasure (“right to be forgotten“)

    You have the right to demand that the controller erase personal data concerning you without undue delay, and we are obligated to erase personal data without undue delay where one of the following grounds applies: the personal data are no longer necessary in relation to the purposes for which they were collected or otherwise processed, the data subject withdraws the consent on which the processing is based according to point (a) of Article 6(1), or point (a) of Article 9(2), and there is no other legal ground for the processing, the data subject objects to the processing pursuant to Article 21(1) GDPR and there are no overriding legitimate grounds for the processing, or the data subject objects to the processing pursuant to Article 21(2) GDPR, the personal data have been unlawfully processed, personal data must be erased for compliance with a legal obligation in Union or Member State law to which the controller is subject, the personal data was collected in relation to the offer of information society services referred to in Article 8(1) GDPR. If the controller has made the personal data public and is obliged pursuant to paragraph 1 to erase the personal data, the controller, taking account of available technology and the cost of implementation, shall take reasonable steps, including technical measures, to inform controllers which are processing the personal data that the data subject has requested the erasure by such controllers of any links to, or copy or replication of, that personal data. The right to erasure (“right to be forgotten“) does not apply to the extent that the processing is necessary: to exercise the right of freedom of expression and information, for compliance with a legal obligation which requires processing by Union or Member, State law to which the controller is subject or for the performance of a task carried out in the public interest or in the exercise of official authority vested in the controller, for reasons of public interest in the area of public health in accordance with points (h) and (i) of Article 9(2) as well as Article 9(3) GDPR, for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes in accordance with Article 89(1) GDPR in so far as the right referred to in paragraph 1 is likely to render impossible or seriously impair the achievement of the objectives of that processing; or for the establishment, exercise or defense of legal claims

    7.6 Right to restriction of processing

    You have the right to request that we restrict the processing of your personal data if any of the following conditions apply: the accuracy of the personal data is contested by the data subject, for a period enabling the controller to verify the accuracy of the personal data, the processing is unlawful and the data subject opposes the erasure of the personal data and requests the restriction of their use instead, the controller no longer needs the personal data for the purposes of the processing, but the data is required by the data subject for the establishment, exercise or defense of legal claims, or the data subject has objected to processing pursuant to Article 21(1) GDPR pending the verification whether the legitimate grounds of the controller override those of the data subject In the event that processing has been restricted under the aforementioned conditions, this personal data shall – with the exception of storage – only be processed with the data subject’s consent or for the establishment, exercise or defense of legal claims or for the protection of the rights of another natural or legal person or for reasons of important public interest of the Union or of a Member State. In order to exercise the right to restrict processing, the data subject may contact us at any time using the contact details provided above.

    7.7 Right to data portability

    You have the right to receive the personal data concerning you which you have provided to us in a structured, commonly used and machine-readable format and have the right to transmit that data to another controller without hindrance from the controller to which the personal data have been provided, to the extent that: the processing is based on consent pursuant to point (a) of Article 6 (1) or point (a) of Article 9 (2) or on a contract pursuant to point (b) of Article 6 (1) GDPR and the processing is carried out by automated means. In exercising your right to data portability pursuant to paragraph 1, you have the right to have the personal data transmitted directly from one controller to another, to the extent that this is technically feasible. The exercise of the right to data portability does not affect your right to erasure (“right to be forgotten”). That right shall not apply to processing necessary for the performance of a task carried out in the public interest or in the exercise of official authority vested in the controller.

    7.8 Right to object

    You have the right to object, on grounds relating to your particular situation, at any time to processing of personal data which concerns you which is based on point (e) or (f) of Article 6 (1) GDPR, including profiling based on those provisions. If objection is made, the controller will no longer process the personal data unless the controller demonstrates compelling legitimate grounds for the processing which override the interests, rights and freedoms of the data subject or for the establishment, exercise or defense of legal claims. In the event that personal data is processed for direct marketing purposes, you have the right to object at any time to processing of personal data concerning you for such marketing. This also applies to profiling to the extent that it is related to such direct marketing. If you object to processing for direct marketing purposes, your personal data shall no longer be processed for such purposes. Regarding the use of information society services, and notwithstanding Directive 2002/58/EC, you can exercise your right to object by automated means using technical specifications. Where personal data are processed for scientific or historical research purposes or statistical purposes pursuant to Article 89 (1), you, on grounds relating to your particular situation, have the right to object to processing of personal data concerning you, unless the processing is necessary for the performance of a task carried out for reasons of public interest. The right of objection can be exercised at any time by contacting the respective controller.

    7.9 Automated individual decision-making, including profiling

    You have the right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects for you or similarly significantly affects you. This does not apply if the decision: is necessary for entering into, or performance of, a contract between the data subject and a data controller, is authorized by Union or Member State law to which the controller is subject and which also lays down suitable measures to safeguard the data subject’s rights and freedoms and legitimate interests, or is based on the data subject’s explicit consent. The controller shall implement suitable measures to safeguard the data subject’s rights and freedoms and legitimate interests, at least the right to obtain human intervention on the part of the controller, to express his or her point of view and to contest the decision. This right can be exercised by the data subject at any time by contacting the respective controller.

    7.10 Right to lodge a complaint with a supervisory authority

    You also have the right, without prejudice to any other administrative or judicial remedy, to lodge a complaint with a supervisory authority, in particular in the Member State of your habitual residence, place of work or place of the alleged infringement if you as data subject consider that the processing of personal data relating to you infringes this Regulation.

    7.11 Right to effective judicial remedy

    Without prejudice to any other available administrative or judicial remedy, including the right to lodge a complaint with a supervisory authority pursuant to Article 77 GDPR, you have the right to an effective judicial remedy if you consider that your rights under this Regulation have been infringed as a result of the processing of your personal data in breach of this Regulation.

    Submitting requests

    If you have any privacy concerns related to this website and the provided datasets or want to have taken images out of the dataset, fill out the following form.

    \ No newline at end of file diff --git a/projects/index.html b/projects/index.html index a2146055..a4a19626 100644 --- a/projects/index.html +++ b/projects/index.html @@ -1 +1 @@ -Projects | LAION

    LAION

    PROJECTS

    DATASETS

    LAION-400M

    image/text

    Status: Released

    Formerly known as crawling@home (C@H), an openly accessible 400M image-text-pair dataset.

    LAION5B

    image/text

    Status: Released

    A 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-coco

    image/text

    Status: Released

    600M captions generated using BLIP from Laion2B-en.

    Laion translated

    image/text

    Status: Released

    3B translated samples from Laion5B.

    Clip H/14

    image/text

    Status: Released

    The largest open source clip.

    LAION5B High-Res

    image/text

    Status: Released

    A subset of the LAION5B database, with high resolution images over 1024x1024, containing 170 million samples.

    LAION Aesthetics

    image/text

    Status: Released

    A subset of LAION5B that has been estimated by a model trained on top of clip embeddings to contain only aestheticly pleasing images.

    LAION-3D

    3d/image/text

    Status: Started

    An effort to create a large-scale dataset consisting of 3D models and descriptor pairs.

    Audio Dataset

    text/audio

    Status: Started

    An 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 Detection

    image/text

    Contrastive

    Status: Released

    A repository containing datasets to train a watermark classifier.

    MODELS

    Openclip

    image/text

    Contrastive

    Status: released

    An open source implementation of OpenAI's CLIP (Contrastive Language-Image Pre-training).

    DALLE-2 Prior/Decoder

    image/text

    Generative

    Status: Started

    An implementation of DALL-E 2, OpenAI's text-to-image synthesis neural network, in Pytorch.

    ClipCap

    image/text

    Generative

    Status: Released

    Generate text from embedding, using pretrained encoder and language models.

    CLAP

    audio/text

    Contrastive

    Status: Started

    A Contrastive Language-Audio Pretraining model, like CLIP, for audio.

    Video CLIP

    video/text

    Contrastive

    Status: Planning

    A contrastive language pretraining model for videos.

    Multilingual-CLIP

    image/text

    Contrastive

    Status: In progress

    An implementation of OpenAI's CLIP text encoders for any language.

    NSFW Detection

    image/text

    Contrastive

    Status: Released

    A detector for not safe for work content within images using CLIP.

    Electric Sheep

    image/text/audio/video

    Contrastive/Generative

    Status: Started

    Train Contrastive and generative models on all modalities.
    \ No newline at end of file +Projects | LAION

    LAION

    PROJECTS

    DATASETS

    LAION-400M

    image/text

    Status: Released

    Formerly known as crawling@home (C@H), an openly accessible 400M image-text-pair dataset.

    LAION5B

    image/text

    Status: Released

    A 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-coco

    image/text

    Status: Released

    600M captions generated using BLIP from Laion2B-en.

    Laion translated

    image/text

    Status: Released

    3B translated samples from Laion5B.

    Clip H/14

    image/text

    Status: Released

    The largest open source clip.

    LAION5B High-Res

    image/text

    Status: Released

    A subset of the LAION5B database, with high resolution images over 1024x1024, containing 170 million samples.

    LAION Aesthetics

    image/text

    Status: Released

    A subset of LAION5B that has been estimated by a model trained on top of clip embeddings to contain only aestheticly pleasing images.

    LAION-3D

    3d/image/text

    Status: Started

    An effort to create a large-scale dataset consisting of 3D models and descriptor pairs.

    Audio Dataset

    text/audio

    Status: Started

    An 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 Detection

    image/text

    Contrastive

    Status: Released

    A repository containing datasets to train a watermark classifier.

    MODELS

    Openclip

    image/text

    Contrastive

    Status: released

    An open source implementation of OpenAI's CLIP (Contrastive Language-Image Pre-training).

    DALLE-2 Prior/Decoder

    image/text

    Generative

    Status: Started

    An implementation of DALL-E 2, OpenAI's text-to-image synthesis neural network, in Pytorch.

    ClipCap

    image/text

    Generative

    Status: Released

    Generate text from embedding, using pretrained encoder and language models.

    CLAP

    audio/text

    Contrastive

    Status: Started

    A Contrastive Language-Audio Pretraining model, like CLIP, for audio.

    Video CLIP

    video/text

    Contrastive

    Status: Planning

    A contrastive language pretraining model for videos.

    Multilingual-CLIP

    image/text

    Contrastive

    Status: In progress

    An implementation of OpenAI's CLIP text encoders for any language.

    NSFW Detection

    image/text

    Contrastive

    Status: Released

    A detector for not safe for work content within images using CLIP.

    Electric Sheep

    image/text/audio/video

    Contrastive/Generative

    Status: Started

    Train Contrastive and generative models on all modalities.
    \ No newline at end of file diff --git a/team/index.html b/team/index.html index abb2f0eb..c074d412 100644 --- a/team/index.html +++ b/team/index.html @@ -1 +1 @@ -Team | LAION

    LAION

    TEAM

    Christoph Schuhmann

    Organizational Lead / Founder

    Master in Physics & Computer Science. Producer of Schools of Trust.

    Jenia Jitsev

    Scientific Lead / Founder

    Senior 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 Vencu

    Engineering Lead / Founder

    AI, automation, and electronics engineer with 28 years of experience in the industry.

    Romain Beaumont

    Open source contributor

    I like to apply scale and deep learning to build AI apps and models.

    Robert Kaczmarczyk

    Community / Medical / Operational Lead / Founder

    Technical University of Munich. 3 years experience in epidemiological research.

    Theo Coombes

    Founding Member

    Programmer with a great passion for big data and machine learning.

    Mehdi Cherti

    Founding Member

    Researcher on Machine Learning and Deep Generative Models. PhD in Computer Science.

    Irina Rish

    Member, senior researcher.

    Professor, Computer Science and Operations Research, Université de Montréal.Strong interest in Scaling Laws for ML

    Aran Komatzusaki

    Member. PhD student of Machine Learning at GaTech

    Interest in RL, representation learning and generative modeling in general

    Robin Rombach

    Member. Stable Diffusion Trainer

    AI researcher with a focus on deep generative models. Author of VQGAN, Latent Diffusion, Stable Diffusion.

    Katherine Crowson

    Member

    AI researcher working on generative models. Generative artist

    Marianna Nezhurina

    Member, core researcher multi-modal learning. PhD student JSC/Tuebingen

    Coder and ML researcher with strong interest in multi-modal datasets and learning.

    Ludwig Schmidt

    Member, senior researcher. Professor in computer science at the University of Washington.

    Research on datasets, reliable generalization, and large models.

    Hilde Kuehne

    Member, senior researcher. Professor in computer science at the University of Bonn, affiliated Llamar Institute & MIT-IBM Watson AI Lab

    Research on Multimodal Self-supervised Learning & Video Understanding.

    Rio Yokota

    Member, 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-Kun

    Member. 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 Rampas

    Technische Hochschule Ingolstadt

    AI researcher, passionate about generative models as means to enhance human creative processes. Author of Paella & Würstchen.

    Pablo Pernias

    Machine Learning Research en Disney Parks

    AI researcher, passionate about generative models as means to enhance human creative processes. Author of Paella & Würstchen.

    Huu Nguyen

    Head of safety policy. Founder and CEO of ontocord.ai

    Computer scientist and lawyer with over 15 years of experience and advocate for the human rights of equal access to scientific advancement and education.

    Björn Plüster

    University of Hamburg, DiscoResearch

    AI researcher with a focus on multilingual LLMs and specialized finetuning.

    Aarush Katta

    Founding Member

    Programmer with lots of enthusiasm for AI & ML.

    Jan Ebert

    Founding Member

    Software engineer and researcher, responsible for scaling up deep learning. Helmholtz AI.

    \ No newline at end of file +Team | LAION

    LAION

    TEAM

    Christoph Schuhmann

    Organizational Lead / Founder

    Master in Physics & Computer Science. Producer of Schools of Trust.

    Jenia Jitsev

    Scientific Lead / Founder

    Senior 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 Vencu

    Engineering Lead / Founder

    AI, automation, and electronics engineer with 28 years of experience in the industry.

    Romain Beaumont

    Open source contributor

    I like to apply scale and deep learning to build AI apps and models.

    Robert Kaczmarczyk

    Community / Medical / Operational Lead / Founder

    Technical University of Munich. 3 years experience in epidemiological research.

    Theo Coombes

    Founding Member

    Programmer with a great passion for big data and machine learning.

    Mehdi Cherti

    Founding Member

    Researcher on Machine Learning and Deep Generative Models. PhD in Computer Science.

    Irina Rish

    Member, senior researcher.

    Professor, Computer Science and Operations Research, Université de Montréal.Strong interest in Scaling Laws for ML

    Aran Komatzusaki

    Member. PhD student of Machine Learning at GaTech

    Interest in RL, representation learning and generative modeling in general

    Robin Rombach

    Member. Stable Diffusion Trainer

    AI researcher with a focus on deep generative models. Author of VQGAN, Latent Diffusion, Stable Diffusion.

    Katherine Crowson

    Member

    AI researcher working on generative models. Generative artist

    Marianna Nezhurina

    Member, core researcher multi-modal learning. PhD student JSC/Tuebingen

    Coder and ML researcher with strong interest in multi-modal datasets and learning.

    Ludwig Schmidt

    Member, senior researcher. Professor in computer science at the University of Washington.

    Research on datasets, reliable generalization, and large models.

    Hilde Kuehne

    Member, senior researcher. Professor in computer science at the University of Bonn, affiliated Llamar Institute & MIT-IBM Watson AI Lab

    Research on Multimodal Self-supervised Learning & Video Understanding.

    Rio Yokota

    Member, 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-Kun

    Member. 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 Rampas

    Technische Hochschule Ingolstadt

    AI researcher, passionate about generative models as means to enhance human creative processes. Author of Paella & Würstchen.

    Pablo Pernias

    Machine Learning Research en Disney Parks

    AI researcher, passionate about generative models as means to enhance human creative processes. Author of Paella & Würstchen.

    Huu Nguyen

    Head of safety policy. Founder and CEO of ontocord.ai

    Computer scientist and lawyer with over 15 years of experience and advocate for the human rights of equal access to scientific advancement and education.

    Björn Plüster

    University of Hamburg, DiscoResearch

    AI researcher with a focus on multilingual LLMs and specialized finetuning.

    Aarush Katta

    Founding Member

    Programmer with lots of enthusiasm for AI & ML.

    Jan Ebert

    Founding Member

    Software engineer and researcher, responsible for scaling up deep learning. Helmholtz AI.

    \ No newline at end of file