<a href=\"https://cdn-uploads.huggingface.co/production/uploads/649176436cadae13f22d014b/eJv58t24-XaYCzWu1s0EZ.png\" rel=\"nofollow\"><img src=\"https://cdn-uploads.huggingface.co/production/uploads/649176436cadae13f22d014b/eJv58t24-XaYCzWu1s0EZ.png\" alt=\"screenshot-20260818-173210\"></a></p>\n","updatedAt":"2026-08-18T09:32:35.654Z","author":{"_id":"649176436cadae13f22d014b","avatarUrl":"/avatars/fa94a8806499a5bd532ca7030e904cd1.svg","fullname":"Higher","name":"HigherHu","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.41390764713287354},"editors":["HigherHu"],"editorAvatarUrls":["/avatars/fa94a8806499a5bd532ca7030e904cd1.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2608.16328","authors":[{"_id":"6a842628b153becad1677116","name":"Feng Xie","hidden":false},{"_id":"6a842628b153becad1677117","name":"Jiagao Hu","hidden":false},{"_id":"6a842628b153becad1677118","name":"Fuhao Li","hidden":false},{"_id":"6a842628b153becad1677119","name":"Zepeng Wang","hidden":false},{"_id":"6a842628b153becad167711a","name":"Yuxuan Chen","hidden":false},{"_id":"6a842628b153becad167711b","name":"Dahua Gao","hidden":false},{"_id":"6a842628b153becad167711c","name":"Fei Wang","hidden":false},{"_id":"6a842628b153becad167711d","name":"Daiguo Zhou","hidden":false}],"publishedAt":"2026-08-17T00:00:00.000Z","submittedOnDailyAt":"2026-08-18T00:00:00.000Z","title":"GRNEdit: Efficient General Video Editing from a New Binary-Evidence Perspective in Generative Refinement Networks","submittedOnDailyBy":{"_id":"649176436cadae13f22d014b","avatarUrl":"/avatars/fa94a8806499a5bd532ca7030e904cd1.svg","isPro":false,"fullname":"Higher","user":"HigherHu","type":"user","name":"HigherHu"},"summary":"Instruction-based general video editing seeks to unify diverse editing operations within a single, intuitive interface. Existing approaches often rely on resource-intensive conditioning, using either heavyweight branches or costly source concatenation. Is there any efficient way to model editing intent? Thus, we introduce GRNEdit, a lightweight two-stage framework. GRN inspires our approach by encoding visual semantics through combinations of bits. Through task-specific fine-tuning, we take this representation further and recast editing semantics as local retain-or-flip decisions over individual bits. Source information is consequently modeled as coordinate-wise evidence supporting the observed binary states, while the GRN backbone remains responsible for resolving their global composition into coherent generative semantics. In Stage I, a compact encoder translates discrete source codes into continuous evidence signals, which GRN assimilates throughout binary refinement. Inspired by null-prompt training for classifier-free guidance, we further assign the null condition an editing-specific meaning: an empty instruction denotes no edit and is supervised through source reconstruction. This identity pathway not only implicitly strengthens evidence utilization and content preservation in Stage I, but also produces a source-preserving state in the same representation space as the edited state. Stage II can therefore directly compare each edited state with its source-preserving counterpart and use their discrepancy to revise unresolved target-bit decisions. Trained on only 0.6M pairs with less than 3\\% conditioning parameters, GRNEdit-2B and GRNEdit-8B achieve scores of 4.03 and 4.18 on OpenVE-Bench. The 2B model outperforms multiple 14B open-source editors, while the 8B model performs on par with leading open-source editors.","upvotes":5,"discussionId":"6a842628b153becad167711e","projectPage":"https://foxerity.github.io/GRNEdit/","githubRepo":"https://github.com/Foxerity/GRNEdit","githubRepoAddedBy":"user","ai_summary":"GRNEdit is a lightweight two-stage framework that models video editing intent via binary semantic decisions and source evidence, achieving strong results with minimal parameters.","ai_keywords":["GRNEdit","binary refinement","coordinate-wise evidence","null-prompt training","classifier-free guidance","identity pathway","source-preserving state","OpenVE-Bench"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":4},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"649176436cadae13f22d014b","avatarUrl":"/avatars/fa94a8806499a5bd532ca7030e904cd1.svg","isPro":false,"fullname":"Higher","user":"HigherHu","type":"user"},{"_id":"66447562e7ca43b97c714cb1","avatarUrl":"/avatars/dc0869849b52650fe4ac7a2c6121a91a.svg","isPro":false,"fullname":"wangzepeng","user":"codingwzp","type":"user"},{"_id":"66d518c48da15c51511635e2","avatarUrl":"/avatars/1446db935bf0f6df45c1316d14501c62.svg","isPro":false,"fullname":"fh","user":"lifuhao","type":"user"},{"_id":"66f6be03b598f4a8b6da9cee","avatarUrl":"/avatars/5f63dafdbb78607a0f0eef99e8daba1a.svg","isPro":false,"fullname":"xf","user":"debugg","type":"user"},{"_id":"6603f231b4344a2b07d5bc05","avatarUrl":"/avatars/da5d570d1e5083a711cdfeca0c6afbad.svg","isPro":false,"fullname":"Ruoyu Wang","user":"dwawayu","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2608/2608.16328.md","query":{}}">
GRNEdit: Efficient General Video Editing from a New Binary-Evidence Perspective in Generative Refinement Networks
Published on Aug 17
· Submitted by Higher on Aug 18 Abstract
GRNEdit is a lightweight two-stage framework that models video editing intent via binary semantic decisions and source evidence, achieving strong results with minimal parameters.
Instruction-based general video editing seeks to unify diverse editing operations within a single, intuitive interface. Existing approaches often rely on resource-intensive conditioning, using either heavyweight branches or costly source concatenation. Is there any efficient way to model editing intent? Thus, we introduce GRNEdit, a lightweight two-stage framework. GRN inspires our approach by encoding visual semantics through combinations of bits. Through task-specific fine-tuning, we take this representation further and recast editing semantics as local retain-or-flip decisions over individual bits. Source information is consequently modeled as coordinate-wise evidence supporting the observed binary states, while the GRN backbone remains responsible for resolving their global composition into coherent generative semantics. In Stage I, a compact encoder translates discrete source codes into continuous evidence signals, which GRN assimilates throughout binary refinement. Inspired by null-prompt training for classifier-free guidance, we further assign the null condition an editing-specific meaning: an empty instruction denotes no edit and is supervised through source reconstruction. This identity pathway not only implicitly strengthens evidence utilization and content preservation in Stage I, but also produces a source-preserving state in the same representation space as the edited state. Stage II can therefore directly compare each edited state with its source-preserving counterpart and use their discrepancy to revise unresolved target-bit decisions. Trained on only 0.6M pairs with less than 3\% conditioning parameters, GRNEdit-2B and GRNEdit-8B achieve scores of 4.03 and 4.18 on OpenVE-Bench. The 2B model outperforms multiple 14B open-source editors, while the 8B model performs on par with leading open-source editors.
Community
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2608.16328 in a model README.md to link it from this page.
Cite arxiv.org/abs/2608.16328 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2608.16328 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.