[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-scion-trains-image-generation-and-vision-understanding-at-once":10,"sections":34},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":24,"tags":25,"sources":29,"feedback":33,"feedback_at":22,"cost_usd":33,"total_tokens":33},10934,"scion-trains-image-generation-and-vision-understanding-at-once","SCION Trains Image Generation and Vision Understanding at Once","A single self-supervised model now generates images and learns visual representations together, no labeled data or pretrained encoders required.","A new research model learns to draw and to see without ever being told what's in a picture.\n\nResearchers built a single model called SCION that handles image generation and self-supervised representation learning together, with no labels and no separately pretrained encoder. One pixel-space encoder does both jobs: a shared learned embedding plus a contrastive loss train its [CLS] token for representation learning, while the same encoder's own [CLS] output conditions a decoder that reconstructs images for generation, with gradient-norm balancing and stop-gradients keeping the two mismatched objectives from clashing during joint training. On the ImageNet 256x256 benchmark, SCION's smaller JiT-B version scored 8.92 FID unguided, beating iREPA (46.44), which aligns to a pretrained DINOv2 encoder, and RCG (14.27), which conditions on one. The larger JiT-L version pushed that to 5.89 FID unguided and 3.47 with guidance, ahead of RCG's comparable 6.24.\n\nThat matters because most strong image generators quietly lean on someone else's labeled dataset or a pretrained vision encoder bolted on for semantic grounding. Folding both jobs into one self-supervised loop removes that dependency and narrows a real chunk of the quality gap to those label- and encoder-assisted methods, echoing the same self-supervised shift that reshaped language models before it reached vision.\n\nIt's one FID score on one benchmark, not a shipped product, and plenty of elegant self-supervised tricks have quietly stalled right there before.","[\"ai\",\"generative-ai\",\"self-supervised-learning\",\"computer-vision\"]","2026-10-09T04:00:00.000Z","2026-10-09T22:22:23.632Z","2026-10-09T22:22:28.622Z","published",null,[],"ai",[24,26,27,28],"generative-ai","self-supervised-learning","computer-vision",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.05711",0,{"sections":35},[36,39,43,48,53,57,61,66,71,76,81,86,91,96],{"name":37,"slug":24,"count":38,"latest_published_at":18},"AI",6708,{"name":40,"slug":41,"count":42,"latest_published_at":18},"Security","security",931,{"name":44,"slug":45,"count":46,"latest_published_at":47},"Policy","policy",486,"2026-10-08T22:40:11.000Z",{"name":49,"slug":50,"count":51,"latest_published_at":52},"Deals","deals",474,"2026-10-08T22:00:00.000Z",{"name":54,"slug":55,"count":56,"latest_published_at":18},"Hardware","hardware",231,{"name":58,"slug":59,"count":60,"latest_published_at":18},"Science","science",192,{"name":62,"slug":63,"count":64,"latest_published_at":65},"Consumer Tech","consumer-tech",181,"2026-10-08T23:26:35.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Startups","startups",117,"2026-10-08T16:45:00.000Z",{"name":72,"slug":73,"count":74,"latest_published_at":75},"Software","software",114,"2026-10-08T17:57:01.000Z",{"name":77,"slug":78,"count":79,"latest_published_at":80},"Dev Tools","dev-tools",105,"2026-10-07T16:59:11.000Z",{"name":82,"slug":83,"count":84,"latest_published_at":85},"General","general",66,"2026-10-09T04:46:11.000Z",{"name":87,"slug":88,"count":89,"latest_published_at":90},"Gaming","gaming",58,"2026-10-08T20:08:45.000Z",{"name":92,"slug":93,"count":94,"latest_published_at":95},"Reviews","reviews",34,"2026-10-08T14:00:22.000Z",{"name":97,"slug":98,"count":99,"latest_published_at":100},"How-To","how-to",8,"2026-10-05T09:00:00.000Z"]