[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-82627-en":3,"doc-seo-82627-105":29,"detail-sidebar-cat-0-en-105":90},{"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":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},82627,16904993612988,"Olivia Brown","https://ap-avatar.wpscdn.com/davatar_a8503ba1806abce46bf441b54a3ca4cd",8,"Research & Report","A Single Patch Is Not Enough: Deterministic Fusion of Repair Candidates","Modern LLM coding agents are evaluated with pass@k, while real development selects a single final patch, creating a pass@k-to-pass@1 decision gap. The paper targets postgeneration selection: candidate pools may contain correct fixes, yet systems must decide which patch to present without reusing shared evidence deterministically. PatchFusion proposes deterministic atomic evidence fusion that fuses diff agreement into a repair neighborhood, selects an auditable representative, and applies evidence-constrained fusion to retain repeated edit atoms and prune unsupported parts, without test or model calls at decision time. A benchmark and results quantify large gains.","A Single Patch Is Not Enough: Deterministic Fusion  \nof Repair Candidates  \nBoyang Yang 1 , Xiangliang Hu2,* , Luyao Ren3,* , Yanjun Chen4  \nBach Le5 , Tegawend F. Bissyand6 , Haoye Tian7,†  \n1Yanshan University; 2Beijing Guoyan Network Data Technology Co., Ltd.  \n3Peking University; 4 Google; 5University of Melbourne  \n6University of Luxembourg; 7Aalto University  \n[yby@ieee.org](yby@ieee.org), [huxiangliang@126.com](huxiangliang@126.com), [rly@pku.edu.cn](rly@pku.edu.cn)  \n[yanjunch98@gmail.com](yanjunch98@gmail.com), [bach.le@unimelb.edu.au](bach.le@unimelb.edu.au), [tegawende.bissyande@uni.lu](tegawende.bissyande@uni.lu)  \n[tianhaoyemail@gmail.com](tianhaoyemail@gmail.com)  \n*Equal contribution. †Corresponding author.  \narXiv :2607 .0 1597v 1 [ cs . SE] 2 Jul 2026  \nAbstract—Modern LLM coding agents are commonly evaluated using pass@k, but developers typically apply a single final patch in real-world settings. This pass@k-to-pass@1 gap is a postgeneration problem: a candidate patch pool may contain a correct patch, but the system must decide which one to suggest to developers. Existing post-generation approaches mainly rank whole candidates, filter them with tests, or query an LLM judge, but none deterministically reuse shared edit-atom evidence to both select and construct the final patch. Thus, we propose PatchFusion, a deterministic atomic evidence fusion approach for candidate patches that consults no test outcome at decision time. PatchFusion first fuses whole-diff agreement into a repair neighborhood, selectsan auditable representative, and then applies evidence-constrained fusion (ECF) to retain repeated edit atoms and prune unsupported parts. To evaluate this setting, we build PatchFuseBench, a fixed-pool benchmark covering SWE-bench Verified, SWE-bench Multilingual, and Defects4J candidate patches.  \nOn PatchFuseBench, PatchFusion solves 426/500 bugs on SWEbench Verified and 236/300 on SWE-bench Multilingual, and reaches 87/371 plausible patches on Defects4J, outperforming every matched candidate-pool selector on all three. PatchFusion recovers 41 and 27 bugs that no single source solves (30 and 18 more over the best single source), repairs that require combining evidence across candidates rather than choosing one patch. It decides in only 3.28 ms per bug, consulting no test or model at decision time. Ablation studies show that ECF adds +5/+6/+9 solved bugs by recovering in-pool repairs that selection misses, with no observed regression, and that PatchFusion’s gains remainstable as candidate pools are resampled. On these complementary multi-source pools, cross-candidate evidence recovers more correct patches than the test-based and LLM-based selectors we evaluate, at orders-of-magnitude lower cost, reaching within 96.2% and 89.7% of the candidate-reachable ceiling on the two SWE-bench benchmarks.  \nI. INTRODUCTION  \nLLM coding agents have improved rapidly and now resolve a large and growing share of real-world software issues, from bug fixes to repository-level edits [1]–[3] . However, behind each reported result, there is not one answer but many: a coding agent samples k candidate patches per bug and is scored by pass@k, which counts a bug as fixed when any of the k samples is correct [4] . A developer applies only a single pass@1 patch, because validating and reviewing every  \ncandidate is budgeted work [5], [6] that prior repair-targeted studies also quantify [7], [8] . Because pass@k can far exceed pass@1, repairs that already sit in the sampled pool are lost the moment one patch must be chosen [9] . This loss grows once the pool spans systems: on SWE-bench [10], different agents and models fix overlapping but distinct bugs, and combining their patches covers far more than the best single system [11]–[13], reaching 443 of 500 against 396 on Verified and 263 of 300 against 218 across 7 models of a multilingual pool. A fixed candidate pool therefore already holds many more repairs than the one patch a developer ","cbCaitclYH87NRQo","https://ap.wps.com/l/cbCaitclYH87NRQo","pdf",504300,1,12,"English","en",105,"# Introduction\n## Pass@k-to-pass@1 Gap\n## Post-generation Repair Selection Challenges\n## Limitations of Whole-Candidate Scoring","[{\"question\":\"What problem does the paper address regarding pass@k evaluation?\",\"answer\":\"It highlights a pass@k-to-pass@1 gap: pass@k deems a bug fixed if any sampled candidate patch is correct, but real use chooses only one final patch, risking loss of fixes present in the candidate pool.\"},{\"question\":\"How does PatchFusion construct the final patch?\",\"answer\":\"PatchFusion deterministically fuses shared edit evidence across candidates: it first fuses whole-diff agreement into a repair neighborhood, selects an auditable representative, then applies evidence-constrained fusion to keep repeated edit atoms and remove unsupported parts.\"},{\"question\":\"What does the evaluation on PatchFuseBench show?\",\"answer\":\"PatchFusion achieves higher bug-solving counts on SWE-bench Verified, SWE-bench Multilingual, and Defects4J, recovers fixes that single-source methods miss, and makes decisions quickly (about 3.28 ms per bug) without consulting tests or models during decision time.\"}]",1784181897,30,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":85,"head_meta":87,"extra_data":89,"updated_unix":27},"a-single-patch-is-not-enough-deterministic-fusion-of-repair-candidates","",{"@graph":35,"@context":84},[36,53,67],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/a-single-patch-is-not-enough-deterministic-fusion-of-repair-candidates/82627/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":61,"encodingFormat":60,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-16",true,{"@type":64,"interactionType":65,"userInteractionCount":4},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"What problem does the paper address regarding pass@k evaluation?","Question",{"text":74,"@type":75},"It highlights a pass@k-to-pass@1 gap: pass@k deems a bug fixed if any sampled candidate patch is correct, but real use chooses only one final patch, risking loss of fixes present in the candidate pool.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"How does PatchFusion construct the final patch?",{"text":79,"@type":75},"PatchFusion deterministically fuses shared edit evidence across candidates: it first fuses whole-diff agreement into a repair neighborhood, selects an auditable representative, then applies evidence-constrained fusion to keep repeated edit atoms and remove unsupported parts.",{"name":81,"@type":72,"acceptedAnswer":82},"What does the evaluation on PatchFuseBench show?",{"text":83,"@type":75},"PatchFusion achieves higher bug-solving counts on SWE-bench Verified, SWE-bench Multilingual, and Defects4J, recovers fixes that single-source methods miss, and makes decisions quickly (about 3.28 ms per bug) without consulting tests or models during decision time.","https://schema.org",{"og:url":51,"og:type":86,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":88,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":91},[92,96,100,104,109,114,119,121,126,129,133],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":105,"doc_module":4,"doc_module_name":45,"category_name":106,"show_sort_weight":107,"slug":108},5,"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":28,"slug":120},"research-report",{"id":122,"doc_module":4,"doc_module_name":45,"category_name":123,"show_sort_weight":124,"slug":125},9,"Religion & Spirituality",20,"religion-spirituality",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":127,"show_sort_weight":124,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":45,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":45,"category_name":135,"show_sort_weight":105,"slug":136},19,"General","general"]