FlowBender enables closed-loop flow matching by learning to correct its own errors, surpassing standard supervised and guidance methods in 2D and 3D conditional generation tasks.<br>Project page: <a href=\"https://flow-bender.github.io/\" rel=\"nofollow\">https://flow-bender.github.io/</a></p>\n","updatedAt":"2026-06-19T07:34:37.514Z","author":{"_id":"672b303918685a38713dd8f1","avatarUrl":"/avatars/800802127af17be6370bf7f65b4a0cc9.svg","fullname":"Daniel Gilo","name":"danielgilo","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.7978220582008362},"editors":["danielgilo"],"editorAvatarUrls":["/avatars/800802127af17be6370bf7f65b4a0cc9.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.20404","authors":[{"_id":"6a34ef8a4c5c5e0d69bf1dad","name":"Daniel Gilo","hidden":false},{"_id":"6a34ef8a4c5c5e0d69bf1dae","name":"Sven Elflein","hidden":false},{"_id":"6a34ef8a4c5c5e0d69bf1daf","name":"Ido Sobol","hidden":false},{"_id":"6a34ef8a4c5c5e0d69bf1db0","name":"Or Litany","hidden":false}],"publishedAt":"2026-06-18T00:00:00.000Z","submittedOnDailyAt":"2026-06-19T00:00:00.000Z","title":"FlowBender: Feedback-Aware Training for Self-Correcting Conditional Flows","submittedOnDailyBy":{"_id":"672b303918685a38713dd8f1","avatarUrl":"/avatars/800802127af17be6370bf7f65b4a0cc9.svg","isPro":false,"fullname":"Daniel Gilo","user":"danielgilo","type":"user","name":"danielgilo"},"summary":"Conditional diffusion and flow models routinely fail to satisfy the very constraints that define their task. For instance, a depth-conditioned model often produces images whose re-extracted depth disagrees with the input, even though the forward operator--the depth predictor defining the constraint--is available during both training and inference. Existing approaches generally fall into two categories: supervised models that treat the conditioning signal as a static cue and ignore alignment information at inference, and guidance-based methods that consult it through hand-tuned linear updates, typically trading fidelity to the condition against the plausibility of the generated sample. We argue that the fundamental gap in both paradigms is that the model is never trained to utilize its own alignment error. We introduce FlowBender, a closed-loop framework that treats this error as a first-class input, training the network to learn a correction policy conditioned on inference-time feedback. At each step, an unguided look-ahead pass estimates the clean signal, a task-specific deviation is computed via the forward operator, and a refinement pass consumes this signal to produce a corrected velocity. We propose several variants of FlowBender, including a gradient-based formulation for differentiable operators and a zero-order variant for non-differentiable settings such as JPEG compression. For efficient sampling, we introduce a prior-step shortcut that enables closed-loop correction at a minimal additional computational cost. Across image-to-image translation, restoration, and 3D mesh texturing, FlowBender consistently outperforms standard supervised baselines, alignment-loss-augmented training, and state-of-the-art inference-time guidance, improving fidelity and plausibility simultaneously rather than trading them against each other. Project page: https://flow-bender.github.io/","upvotes":4,"discussionId":"6a34ef8a4c5c5e0d69bf1db1","projectPage":"https://flow-bender.github.io/","ai_summary":"FlowBender is a closed-loop framework that addresses constraint satisfaction in diffusion and flow models by training networks to correct alignment errors using inference-time feedback, outperforming traditional supervised and guidance-based approaches across multiple tasks.","ai_keywords":["conditional diffusion models","flow models","alignment error","closed-loop framework","unguided look-ahead pass","refinement pass","gradient-based formulation","zero-order variant","inference-time guidance","image-to-image translation","restoration","3D mesh texturing"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","organization":{"_id":"6393322be2364bc1eea56e45","name":"Technion","fullname":"Technion Israel institute of technology","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/1670591001944-63926124526c29d5b5011374.jpeg"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"672b303918685a38713dd8f1","avatarUrl":"/avatars/800802127af17be6370bf7f65b4a0cc9.svg","isPro":false,"fullname":"Daniel Gilo","user":"danielgilo","type":"user"},{"_id":"676d699f6f93172ba941a382","avatarUrl":"/avatars/15dddb7b577bb202646dbab3f249d9b2.svg","isPro":false,"fullname":"Tomer Borreda","user":"TomerBo","type":"user"},{"_id":"66476eff9e82dfd4a370a572","avatarUrl":"/avatars/15715bad8057962d0102f52fb2b9d88b.svg","isPro":false,"fullname":"Ido Sobol","user":"IdoSobol","type":"user"},{"_id":"692da82ab884a1ba56270557","avatarUrl":"/avatars/e98409da85032ad2092b77f9fceee784.svg","isPro":false,"fullname":"Sven Elflein","user":"sven-el","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"6393322be2364bc1eea56e45","name":"Technion","fullname":"Technion Israel institute of technology","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/1670591001944-63926124526c29d5b5011374.jpeg"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2606/2606.20404.md","query":{}}">
FlowBender: Feedback-Aware Training for Self-Correcting Conditional Flows
Abstract
FlowBender is a closed-loop framework that addresses constraint satisfaction in diffusion and flow models by training networks to correct alignment errors using inference-time feedback, outperforming traditional supervised and guidance-based approaches across multiple tasks.
Conditional diffusion and flow models routinely fail to satisfy the very constraints that define their task. For instance, a depth-conditioned model often produces images whose re-extracted depth disagrees with the input, even though the forward operator--the depth predictor defining the constraint--is available during both training and inference. Existing approaches generally fall into two categories: supervised models that treat the conditioning signal as a static cue and ignore alignment information at inference, and guidance-based methods that consult it through hand-tuned linear updates, typically trading fidelity to the condition against the plausibility of the generated sample. We argue that the fundamental gap in both paradigms is that the model is never trained to utilize its own alignment error. We introduce FlowBender, a closed-loop framework that treats this error as a first-class input, training the network to learn a correction policy conditioned on inference-time feedback. At each step, an unguided look-ahead pass estimates the clean signal, a task-specific deviation is computed via the forward operator, and a refinement pass consumes this signal to produce a corrected velocity. We propose several variants of FlowBender, including a gradient-based formulation for differentiable operators and a zero-order variant for non-differentiable settings such as JPEG compression. For efficient sampling, we introduce a prior-step shortcut that enables closed-loop correction at a minimal additional computational cost. Across image-to-image translation, restoration, and 3D mesh texturing, FlowBender consistently outperforms standard supervised baselines, alignment-loss-augmented training, and state-of-the-art inference-time guidance, improving fidelity and plausibility simultaneously rather than trading them against each other. Project page: https://flow-bender.github.io/
Community
FlowBender enables closed-loop flow matching by learning to correct its own errors, surpassing standard supervised and guidance methods in 2D and 3D conditional generation tasks.
Project page: https://flow-bender.github.io/
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/2606.20404 in a model README.md to link it from this page.
Cite arxiv.org/abs/2606.20404 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2606.20404 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.