{"id":40009,"date":"2025-07-23T11:53:12","date_gmt":"2025-07-23T03:53:12","guid":{"rendered":"https:\/\/www.1ai.net\/?p=40009"},"modified":"2025-07-23T11:53:12","modified_gmt":"2025-07-23T03:53:12","slug":"%e5%ad%97%e8%8a%82%e8%b7%b3%e5%8a%a8%e5%8f%91%e5%b8%83%e9%80%9a%e7%94%a8%e6%9c%ba%e5%99%a8%e4%ba%ba%e6%a8%a1%e5%9e%8b%e3%80%8cseed-gr-3%e3%80%8d","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/40009.html","title":{"rendered":"ByteDance Releases General Purpose Robot Model 'Seed GR-3'"},"content":{"rendered":"<p>July 22<a href=\"https:\/\/www.1ai.net\/en\/tag\/%e5%ad%97%e8%8a%82%e8%b7%b3%e5%8a%a8\" title=\"[View articles tagged with [bytejump]]\" target=\"_blank\" >ByteDance<\/a> The Seed team has announced its latest result: a generalizable, complex operational task that supports long sequences of<a href=\"https:\/\/www.1ai.net\/en\/tag\/%e6%9c%ba%e5%99%a8%e4%ba%ba\" title=\"[Sees articles with [robots] labels]\" target=\"_blank\" >robot<\/a>Operation of the large model \"Seed <a href=\"https:\/\/www.1ai.net\/en\/tag\/gr-3\" title=\"[SEE ARTICLES WITH [GR-3] LABELS]\" target=\"_blank\" >GR-3<\/a>&quot;.<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-40010\" title=\"358dae71j00szu2mz001yd000u000gwm\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2025\/07\/358dae71j00szu2mz001yd000u000gwm.jpg\" alt=\"358dae71j00szu2mz001yd000u000gwm\" width=\"1080\" height=\"608\" \/><\/p>\n<p>GR-3 IS DESCRIBED AS A LARGE-SCALE VISUAL-LANGUAGE-ACTION (VLA) MODEL. IT DEMONSTRATES A BETTER GENERALIZATION OF NEW OBJECTS, NEW ENVIRONMENTS AND NEW COMMANDS WITH ABSTRACT CONCEPTS. IN ADDITION, GR-3 SUPPORTS THE EFFICIENT FINE-TUNING OF A SMALL AMOUNT OF HUMAN TRAJECTORIES DATA THAT CAN RAPIDLY AND ECONOMICALLY ADAPT TO NEW TASKS. GR-3 ALSO DISPLAYS ROBUST AND RELIABLE PERFORMANCE IN HANDLING LONG-CYCLE AND SMART MISSIONS, INCLUDING THOSE REQUIRING HAND-TO-HAND AND CHASSIS MOVEMENT\u3002<\/p>\n<p>Seed team presentation, these competencies are derived from a variety of training methods: Using large-scale visual-linguistic data combined training, the collection of user-authorized human trajectory data based on VR devices for efficient fine-tuning and the effective imitation of learning based on robotic trajectory data\u3002<\/p>\n<p>It is worth mentioning that GR-3 performed better in the challenging flexible object manipulation test (specifically tested as a clothes hanging task), and was able to perform the operation of \"threading a hanger into the clothes and then hanging them on a clothesline\". In addition, GR-3 was able to generalize to clothing types not included in the training data.<\/p>\n<p>Additionally, the Seed team has introduced ByteMini, a two-armed mobile robot that is said to combine dexterity and reliability with the integration of GR-3 to perform a wide range of complex tasks.<\/p>","protected":false},"excerpt":{"rendered":"<p>On July 22, the byte beat Seed team released its latest results: a broad robotic operation model \"Seed GR-3\" that can be generalized and supports long series complex operations. GR-3 is described as a large-scale visual-language-action (VLA) model. It demonstrates a better generalization of new objects, new environments and new commands with abstract concepts. In addition, GR-3 supports the efficient fine-tuning of a small amount of human trajectories data that can rapidly and economically adapt to new tasks. GR-3 also displays robust and reliable performance in handling long-cycle and smart missions, including those requiring hand-to-hand and chassis movement. Seed team presentation, these competencies are derived from a variety of training methods:<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[146],"tags":[7252,548,909],"collection":[],"class_list":["post-40009","post","type-post","status-publish","format-standard","hentry","category-news","tag-gr-3","tag-548","tag-909"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/40009","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=40009"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/40009\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=40009"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=40009"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=40009"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=40009"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}