[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84101-en":3,"doc-seo-84101-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},84101,1099514067438,"River Wang","https://ap-avatar.wpscdn.com/avatar/100002539ee87300030?x-image-process=image/resize,m_fixed,w_180,h_180&k=1780474512215547542",8,"Research & Report","Bridging Diffusion Pruning and Step Distillation with Teacher-Aligned Repair","Diffusion models produce high-quality images, yet inference is expensive due to a large denoising network and repeated denoising steps. Existing compression methods typically optimize these costs separately: structured pruning shrinks the network but often requires lengthy post-pruning retraining to restore many-step sampling, while step distillation reduces denoising evaluations but assumes a student already aligns with the teacher’s denoising field. Direct replacement fails after pruning. A short teacher-alignment repair bridge matches the pruned generator to the teacher on noisy real-image latents, enabling subsequent one-step distillation.","BRIDGING DIFFUSION PRUNING AND STEP DISTILLATION WITH  \nTEACHER-ALIGNED REPAIR  \nJincheng Ying, Li Wenlin  \nSchool of Big Data and Artificial Intelligence Guangdong University of Finance and Economics Guangzhou, China [jc_ying@student.gdufe.edu.cn](jc_ying@student.gdufe.edu.cn), [wenlin@gdufe.edu.cn](wenlin@gdufe.edu.cn)  \nMinghui Xu  \nSchool of Computer Science Shandong University Qingdao, China [mhxu@sdu.edu.cn](mhxu@sdu.edu.cn)  \narXiv :2607 .06335v 1 [ cs .CV] 7 Jul 2026  \nYinhao Xiao∗  \nGDUFE-UCM Joint Institute  \nGuangdong University of Finance and Economics  \nGuangzhou, China  \n[20191081@gdufe.edu.cn](20191081@gdufe.edu.cn)  \nABSTRACT  \nDiffusion models generate high-quality images, but their inference cost comes from two sources: large denoising networks and repeated denoising steps. Existing compression pipelines usually attack these costs separately. Pruning reduces the network, but most pruning methods still rely on a long post-pruning retraining stage to recover a many-step sampler. Step distillation reduces the number of denoising steps, but it usually assumes a student that can already follow the teacher well enough to receive useful distillation gradients. This paper asks whether post-pruning retraining can be replaced by step distillation. We find that the direct replacement fails: after pruning an EDM2-XS teacher, starting SiDA from the pruned checkpoint produces unusable samples. We introduce a short teacher-alignment repair stage as a bridge between pruning and step distillation. The bridge matches the pruned generator to the teacher on noisy real-image latents, then hands the repaired checkpoint to one-step distillation. On ImageNet-512, the original EDM2-XS baseline uses 124.713M parameters and 63 network evaluations, reaching an FID of 3.53 . With a suitable distillation objective, our 20% pruned one-step generator uses 98.826M parameters and one network evaluation, reaching an FID of  \n3.12. With 30% pruning, the model uses 88.029M parameters and one network evaluation, with an FID of 4 .26.  \nKeywords diffusion distillation · one-step generation · structured pruning  \n1 Introduction  \nDiffusion models have become a central tool for high-quality image synthesis, including unconditional generation, class-conditional generation, text-to-image generation, editing, and super-resolution [1, 2, 3, 4] . Their cost has two sources. The denoising network is large, and sampling evaluates it repeatedly. On ImageNet-512, the EDM2-XS sampler in our setup uses 63 network evaluations per image.  \nMost efficiency work reduces one of these two costs. Pruning removes weights, channels, blocks, or heads from apretrained diffusion model [5, 6, 7] . Step distillation reduces the number of denoising evaluations, sometimes to one ora few steps [8, 9, 10, 11] . These directions are complementary, but they are usually not connected. A pruned diffusion model is commonly recovered by retraining or fine-tuning the pruned network as a many-step sampler. Step distillation, in contrast, is normally applied to a student that is already compatible with the teacher’s denoising field.  \n∗ Corresponding author.  \nBridging Pruning and Step Distillation  \nTable 1: Replacing post-pruning retraining with step distillation is not a direct plug-in. The pruned checkpoint fails under direct one-step distillation, while the proposed repair bridge makes the same recovery path usable.  \n\n| Model | Repair | NFE | \\#Params | FID (↓) |\n| --- | --- | --- | --- | --- |\n| EDM2-XS teacher [4] | – | 63 | 124.713M | 3.53 |\n| Pruned + direct SiDA | No | 1 | 88.029M | 112.56 (failed) |\n| Ours (20% pruned) | Yes | 1 | 98.826M | 3.12 |\n| Ours (30% pruned) | Yes | 1 | 88.029M | 4.26 |\n\nThis raises a direct question: can the expensive post-pruning retraining stage be replaced by step distillation? If so, pruning would reduce the network size and distillation would reduce the sampling trajectory in the same recovery process. The recovered model would not be a smalle","cbCaikTNDbrpidDQ","https://ap.wps.com/l/cbCaikTNDbrpidDQ","pdf",9504853,4,1,14,"English","en",105,"# Abstract\n# Introduction\n# Related Work\n## Pruning Diffusi","[{\"question\":\"Why can’t post-pruning retraining be directly replaced by one-step distillation?\",\"answer\":\"After structured pruning, the student can be far from the teacher on noisy states where distillation loss is computed. The one-step distillation then receives a damaged initialization, leading to unusable samples.\"},{\"question\":\"What is the proposed teacher-alignment repair bridge?\",\"answer\":\"It introduces a short repair stage that matches the pruned generator to the teacher on noisy real-image latents in the high-noise region used by one-step generation, without running full retraining.\"},{\"question\":\"What improvements are reported on ImageNet-512 for EDM2-XS?\",\"answer\":\"The baseline uses 63 network evaluations with 124.713M parameters and FID 3.53. With suitable distillation, a 20% pruned one-step generator uses 98.826M parameters and 1 evaluation, reaching FID 3.12; with 30% pruning it uses 88.029M parameters and 1 evaluation, reaching FID 4.26.\"}]",1784192796,35,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"bridging-diffusion-pruning-and-step-distillation-with-teacher-aligned-repair","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":20},"https://docshare.wps.com/document/bridging-diffusion-pruning-and-step-distillation-with-teacher-aligned-repair/84101/",{"url":52,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-28","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why can’t post-pruning retraining be directly replaced by one-step distillation?","Question",{"text":75,"@type":76},"After structured pruning, the student can be far from the teacher on noisy states where distillation loss is computed. The one-step distillation then receives a damaged initialization, leading to unusable samples.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What is the proposed teacher-alignment repair bridge?",{"text":80,"@type":76},"It introduces a short repair stage that matches the pruned generator to the teacher on noisy real-image latents in the high-noise region used by one-step generation, without running full retraining.",{"name":82,"@type":73,"acceptedAnswer":83},"What improvements are reported on ImageNet-512 for EDM2-XS?",{"text":84,"@type":76},"The baseline uses 63 network evaluations with 124.713M parameters and FID 3.53. With suitable distillation, a 20% pruned one-step generator uses 98.826M parameters and 1 evaluation, reaching FID 3.12; with 30% pruning it uses 88.029M parameters and 1 evaluation, reaching FID 4.26.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":20,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]