Librarian Bot. I found the following papers similar to this paper.

The following papers were recommended by the Semantic Scholar API

KOALA: Self-Attention Matters in Knowledge Distillation of Latent Diffusion Models for Memory-Efficient and Fast Image Synthesis (2023)
A-SDM: Accelerating Stable Diffusion through Redundancy Removal and Performance Optimization (2023)
LoRA-Enhanced Distillation on Guided Diffusion Models (2023)
SpeedUpNet: A Plug-and-Play Hyper-Network for Accelerating Text-to-Image Diffusion Models (2023)
ECLIPSE: A Resource-Efficient Text-to-Image Prior for Image Generations (2023)

Please give a thumbs up to this comment if you found it helpful!

If you want recommendations for any Paper on Hugging Face checkout this Space

\n","updatedAt":"2024-01-09T09:53:58.145Z","author":{"_id":"63d3e0e8ff1384ce6c5dd17d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1674830754237-63d3e0e8ff1384ce6c5dd17d.jpeg","fullname":"Librarian Bot (Bot)","name":"librarian-bot","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":318,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.6847825050354004},"editors":["librarian-bot"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/1674830754237-63d3e0e8ff1384ce6c5dd17d.jpeg"],"reactions":[],"isReport":false}},{"id":"6664f96c14e005a1e65c1705","author":{"_id":"6186ddf6a7717cb375090c01","avatarUrl":"/avatars/716b6a7d1094c8036b2a8a7b9063e8aa.svg","fullname":"Julien BLANCHON","name":"blanchon","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":176,"isUserFollowing":false},"createdAt":"2024-06-09T00:38:04.000Z","type":"comment","data":{"edited":false,"hidden":false,"latest":{"raw":"# Scaling Down AI: Breakthrough in Efficient Stable Diffusion Models!\n\nhttps://cdn-uploads.huggingface.co/production/uploads/6186ddf6a7717cb375090c01/xbwDkwF9ZN8dNfb7aS1LT.mp4 \n\n## Links 🔗:\n👉 Subscribe: https://www.youtube.com/@Arxflix\n👉 Twitter: https://x.com/arxflix\n👉 LMNT (Partner): https://lmnt.com/\n\n\nBy Arxflix\n![9t4iCUHx_400x400-1.jpg](https://cdn-uploads.huggingface.co/production/uploads/6186ddf6a7717cb375090c01/v4S5zBurs0ouGNwYj1GEd.jpeg)","html":"

\n\t\n\t\t\n\t\n\t\n\t\tScaling Down AI: Breakthrough in Efficient Stable Diffusion Models!\n\t\n

\n\n

\n\t\n\t\t\n\t\n\t\n\t\tLinks 🔗:\n\t\n

👉 Subscribe: https://www.youtube.com/@Arxflix
👉 Twitter: https://x.com/arxflix
👉 LMNT (Partner): https://lmnt.com/

By Arxflix
$\"9t4iCUHx_400x400-1.jpg\"$

\n","updatedAt":"2024-06-09T00:38:04.177Z","author":{"_id":"6186ddf6a7717cb375090c01","avatarUrl":"/avatars/716b6a7d1094c8036b2a8a7b9063e8aa.svg","fullname":"Julien BLANCHON","name":"blanchon","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":176,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.48688939213752747},"editors":["blanchon"],"editorAvatarUrls":["/avatars/716b6a7d1094c8036b2a8a7b9063e8aa.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2401.02677","authors":[{"_id":"659b635821a74316431e14e8","user":{"_id":"62f8ca074588fe31f4361dae","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/62f8ca074588fe31f4361dae/F2k343TPD7KVfW3P26IRs.jpeg","isPro":false,"fullname":"Yatharth Gupta","user":"Warlord-K","type":"user"},"name":"Yatharth Gupta","status":"claimed_verified","statusLastChangedAt":"2024-01-08T08:47:26.552Z","hidden":false},{"_id":"659b635821a74316431e14e9","user":{"_id":"639362039694ec0d025f194d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/639362039694ec0d025f194d/jmQetjDK3ISH_IQb7SBGC.png","isPro":false,"fullname":"Vishnu V Jaddipal","user":"Icar","type":"user"},"name":"Vishnu V. Jaddipal","status":"claimed_verified","statusLastChangedAt":"2024-01-08T08:47:28.507Z","hidden":false},{"_id":"659b635821a74316431e14ea","user":{"_id":"6412b77da26724794c3c82b5","avatarUrl":"/avatars/3f20c2698b29462f1e5231bbaa51478b.svg","isPro":false,"fullname":"Harish Segmind","user":"harishseg","type":"user"},"name":"Harish Prabhala","status":"extracted_confirmed","statusLastChangedAt":"2024-01-08T04:13:21.785Z","hidden":false},{"_id":"659b635821a74316431e14eb","user":{"_id":"5f7fbd813e94f16a85448745","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1649681653581-5f7fbd813e94f16a85448745.jpeg","isPro":true,"fullname":"Sayak Paul","user":"sayakpaul","type":"user"},"name":"Sayak Paul","status":"extracted_confirmed","statusLastChangedAt":"2024-01-08T07:39:54.480Z","hidden":false},{"_id":"659b635821a74316431e14ec","user":{"_id":"5dfcb1aada6d0311fd3d5448","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1584435275418-5dfcb1aada6d0311fd3d5448.jpeg","isPro":false,"fullname":"Patrick von Platen","user":"patrickvonplaten","type":"user"},"name":"Patrick Von Platen","status":"extracted_pending","statusLastChangedAt":"2024-01-08T02:52:12.391Z","hidden":false}],"publishedAt":"2024-01-05T07:21:46.000Z","submittedOnDailyAt":"2024-01-08T00:22:12.414Z","title":"Progressive Knowledge Distillation Of Stable Diffusion XL Using Layer\n Level Loss","submittedOnDailyBy":{"_id":"60f1abe7544c2adfd699860c","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg","isPro":false,"fullname":"AK","user":"akhaliq","type":"user"},"summary":"Stable Diffusion XL (SDXL) has become the best open source text-to-image\nmodel (T2I) for its versatility and top-notch image quality. Efficiently\naddressing the computational demands of SDXL models is crucial for wider reach\nand applicability. In this work, we introduce two scaled-down variants, Segmind\nStable Diffusion (SSD-1B) and Segmind-Vega, with 1.3B and 0.74B parameter\nUNets, respectively, achieved through progressive removal using layer-level\nlosses focusing on reducing the model size while preserving generative quality.\nWe release these models weights at https://hf.co/Segmind. Our methodology\ninvolves the elimination of residual networks and transformer blocks from the\nU-Net structure of SDXL, resulting in significant reductions in parameters, and\nlatency. Our compact models effectively emulate the original SDXL by\ncapitalizing on transferred knowledge, achieving competitive results against\nlarger multi-billion parameter SDXL. Our work underscores the efficacy of\nknowledge distillation coupled with layer-level losses in reducing model size\nwhile preserving the high-quality generative capabilities of SDXL, thus\nfacilitating more accessible deployment in resource-constrained environments.","upvotes":25,"discussionId":"659b635c21a74316431e1618","githubRepo":"https://github.com/segmind/ssd-1b","githubRepoAddedBy":"auto","ai_summary":"Two scaled-down variants of Stable Diffusion XL (SDXL), achieved through layer-level loss-driven pruning, maintain competitive image quality with reduced computational demands.","ai_keywords":["Stable Diffusion XL","text-to-image","T2I","parameter-efficient","U-Net","residual networks","transformer blocks","knowledge distillation","layer-level losses","generative quality","resource-constrained environments"],"githubStars":179},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"620783f24e28382272337ba4","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/620783f24e28382272337ba4/zkUveQPNiDfYjgGhuFErj.jpeg","isPro":false,"fullname":"GuoLiangTang","user":"Tommy930","type":"user"},{"_id":"5f7fbd813e94f16a85448745","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1649681653581-5f7fbd813e94f16a85448745.jpeg","isPro":true,"fullname":"Sayak Paul","user":"sayakpaul","type":"user"},{"_id":"639362039694ec0d025f194d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/639362039694ec0d025f194d/jmQetjDK3ISH_IQb7SBGC.png","isPro":false,"fullname":"Vishnu V Jaddipal","user":"Icar","type":"user"},{"_id":"6412b77da26724794c3c82b5","avatarUrl":"/avatars/3f20c2698b29462f1e5231bbaa51478b.svg","isPro":false,"fullname":"Harish Segmind","user":"harishseg","type":"user"},{"_id":"62039c2d91d53938a643317d","avatarUrl":"/avatars/e869979d907f388526ed6273bb3ba150.svg","isPro":false,"fullname":"Harish Prabhala","user":"harishp","type":"user"},{"_id":"62f8ca074588fe31f4361dae","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/62f8ca074588fe31f4361dae/F2k343TPD7KVfW3P26IRs.jpeg","isPro":false,"fullname":"Yatharth Gupta","user":"Warlord-K","type":"user"},{"_id":"6311bca0ae8896941da24e66","avatarUrl":"/avatars/48de64894fc3c9397e26e4d6da3ff537.svg","isPro":false,"fullname":"Fynn Kröger","user":"fynnkroeger","type":"user"},{"_id":"61fc1baa38ece86f733d84d9","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1643912098555-noauth.jpeg","isPro":false,"fullname":"Anthonny OLIME","user":"Citaman","type":"user"},{"_id":"60d2dc1007da9c17c72708f8","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1624431552569-noauth.jpeg","isPro":false,"fullname":"yuvraj sharma","user":"ysharma","type":"user"},{"_id":"648eb1eb59c4e5c87dc116e0","avatarUrl":"/avatars/c636cea39c2c0937f01398c94ead5dad.svg","isPro":false,"fullname":"fdsqefsgergd","user":"T-representer","type":"user"},{"_id":"5dd96eb166059660ed1ee413","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/NQtzmrDdbG0H8qkZvRyGk.jpeg","isPro":true,"fullname":"Julien Chaumond","user":"julien-c","type":"user"},{"_id":"633e72fc3a17ab61de8cdc5f","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/633e72fc3a17ab61de8cdc5f/5lPKWN1C7AR09YcJo-B6O.png","isPro":false,"fullname":"pip zambo","user":"froilo","type":"user"}],"acceptLanguages":["*"],"dailyPaperRank":0}">

Papers

arxiv:2401.02677

Progressive Knowledge Distillation Of Stable Diffusion XL Using Layer Level Loss

Published on Jan 5, 2024

· Submitted by

AK on Jan 8, 2024

Upvote

Authors:

Yatharth Gupta ,

Vishnu V. Jaddipal ,

Harish Prabhala ,

Sayak Paul ,

Patrick Von Platen

Abstract

Two scaled-down variants of Stable Diffusion XL (SDXL), achieved through layer-level loss-driven pruning, maintain competitive image quality with reduced computational demands.

AI-generated summary

Stable Diffusion XL (SDXL) has become the best open source text-to-image model (T2I) for its versatility and top-notch image quality. Efficiently addressing the computational demands of SDXL models is crucial for wider reach and applicability. In this work, we introduce two scaled-down variants, Segmind Stable Diffusion (SSD-1B) and Segmind-Vega, with 1.3B and 0.74B parameter UNets, respectively, achieved through progressive removal using layer-level losses focusing on reducing the model size while preserving generative quality. We release these models weights at https://hf.co/Segmind. Our methodology involves the elimination of residual networks and transformer blocks from the U-Net structure of SDXL, resulting in significant reductions in parameters, and latency. Our compact models effectively emulate the original SDXL by capitalizing on transferred knowledge, achieving competitive results against larger multi-billion parameter SDXL. Our work underscores the efficacy of knowledge distillation coupled with layer-level losses in reducing model size while preserving the high-quality generative capabilities of SDXL, thus facilitating more accessible deployment in resource-constrained environments.