\n","updatedAt":"2024-06-09T02:55:20.461Z","author":{"_id":"6186ddf6a7717cb375090c01","avatarUrl":"/avatars/716b6a7d1094c8036b2a8a7b9063e8aa.svg","fullname":"Julien BLANCHON","name":"blanchon","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":142}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.4852614998817444},"editors":["blanchon"],"editorAvatarUrls":["/avatars/716b6a7d1094c8036b2a8a7b9063e8aa.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2207.12598","authors":[{"_id":"6411c77f6b75ddced3890cc0","name":"Jonathan Ho","hidden":false},{"_id":"6411c77f6b75ddced3890cc1","name":"Tim Salimans","hidden":false}],"publishedAt":"2022-07-26T01:42:07.000Z","title":"Classifier-Free Diffusion Guidance","summary":"Classifier guidance is a recently introduced method to trade off mode\ncoverage and sample fidelity in conditional diffusion models post training, in\nthe same spirit as low temperature sampling or truncation in other types of\ngenerative models. Classifier guidance combines the score estimate of a\ndiffusion model with the gradient of an image classifier and thereby requires\ntraining an image classifier separate from the diffusion model. It also raises\nthe question of whether guidance can be performed without a classifier. We show\nthat guidance can be indeed performed by a pure generative model without such a\nclassifier: in what we call classifier-free guidance, we jointly train a\nconditional and an unconditional diffusion model, and we combine the resulting\nconditional and unconditional score estimates to attain a trade-off between\nsample quality and diversity similar to that obtained using classifier\nguidance.","upvotes":3,"discussionId":"641192363ea54b1aa7e2f435","ai_summary":"Classifier-free guidance achieves a balance between sample quality and diversity in conditional diffusion models by jointly training both conditional and unconditional models.","ai_keywords":["classifier guidance","conditional diffusion models","score estimate","image classifier","truncation","low temperature sampling","classifier-free guidance","unconditional diffusion models","sample quality","diversity"]},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"655ac762cb17ec19ef82719b","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/655ac762cb17ec19ef82719b/1kDncYrGLYS_2SR8cNdAL.png","isPro":false,"fullname":"Welcome to matlok","user":"matlok","type":"user"},{"_id":"6484c61300f3d63d6c3d6b27","avatarUrl":"/avatars/ce9b99882a65fd2cb983ba71a5ac2473.svg","isPro":false,"fullname":"Aryan V S","user":"a-r-r-o-w","type":"user"},{"_id":"63c6cb6a50cc81901da65e15","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63c6cb6a50cc81901da65e15/t4LN1BPCFlwbSJ9GD9YDd.jpeg","isPro":true,"fullname":"Thรฉo Pomies","user":"theopomies","type":"user"}],"acceptLanguages":["*"]}">
Classifier-free guidance achieves a balance between sample quality and diversity in conditional diffusion models by jointly training both conditional and unconditional models.
AI-generated summary
Classifier guidance is a recently introduced method to trade off mode
coverage and sample fidelity in conditional diffusion models post training, in
the same spirit as low temperature sampling or truncation in other types of
generative models. Classifier guidance combines the score estimate of a
diffusion model with the gradient of an image classifier and thereby requires
training an image classifier separate from the diffusion model. It also raises
the question of whether guidance can be performed without a classifier. We show
that guidance can be indeed performed by a pure generative model without such a
classifier: in what we call classifier-free guidance, we jointly train a
conditional and an unconditional diffusion model, and we combine the resulting
conditional and unconditional score estimates to attain a trade-off between
sample quality and diversity similar to that obtained using classifier
guidance.