[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-267883-105":59,"doc-detail-267883-en":131},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":124,"head_meta":126,"extra_data":128,"updated_unix":130},105,"en","sepo-evidence-grounded-prompt-optimization-via-structural-editing","SEPO: Evidence-grounded Prompt Optimization via Structural Editing","","API-only prompt optimisers are often claimed to be interpretable, yet they typically rewrite prompts as opaque strings, producing traces of full-prompt diffs without local, machine-readable edit meaning. SEPO (Structural, Evidence-grounded Prompt Optimization) introduces a multi-trajectory optimiser that performs local edits on stable typed units and records edit-effect lineage linking structural operations to examples fixed or broken. Across a 14-task held-out suite, it improves macro accuracy over GEPA and matches Pareto efficiency.",{"@graph":69,"@context":123},[70,84,106],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/sepo-evidence-grounded-prompt-optimization-via-structural-editing/267883/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/sepo-evidence-grounded-prompt-optimization-via-structural-editing/267883.png","ImageObject",300,407,{"name":92,"@type":93},"Theodore","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-11-03","2026-09-14",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",12,{"@type":107,"mainEntity":108},"FAQPage",[109,115,119],{"name":110,"@type":111,"acceptedAnswer":112},"Why do existing API-only prompt optimisers lack true interpretability?","Question",{"text":113,"@type":114},"They usually rewrite the entire prompt as an opaque string each iteration, yielding full-prompt diff traces that do not indicate where relevant changes occurred or which examples they helped or harmed.","Answer",{"name":116,"@type":111,"acceptedAnswer":117},"How does SEPO perform prompt optimisation differently?",{"text":118,"@type":114},"SEPO locally edits stable, typed units in a two-layer prompt schema and tracks the realised structural operations, linking each edit to the examples it newly fixes or breaks to guide later calls.",{"name":120,"@type":111,"acceptedAnswer":121},"What performance and efficiency benefits does SEPO show?",{"text":122,"@type":114},"On a 14-task held-out suite, SEPO improves macro accuracy over the best baseline, GEPA, and reaches Pareto frontiers while using fewer optimisation tokens.","https://schema.org",{"og:url":83,"og:type":125,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":127,"canonical":83},"index,follow",{"doc_id":129,"site_id":62},267883,1790065627,{"code":4,"msg":5,"data":132},{"doc_id":129,"user_id":133,"nickname":92,"user_avatar":134,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":140,"language":141,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":142,"faqs":143,"seo_title":144,"seo_description":67,"update_tm":145,"read_time":146,"preview_page_count":4},7971461740886,"https://ap-avatar.wpscdn.com/davatar_3d24733baf745e90a7e4bdd5f77d97b2","SEPO: Evidence-Grounded Prompt Optimization via  \nStructural Editing  \nXiaoyu Ma♠† Haoyue Liu♠† Yiwen Li♠† Jionghao Zhu♠ Zhichao Wang♠ Ye Chen♡ Xiaoying Tang♠♣♢∗  \n♠ School of Science and Engineering, The Chinese University of Hong Kong, Shenzhen  \n♣ Shenzhen Future Network of Intelligence Institute (FNiI-Shenzhen)♢ Guangdong Provincial Key Laboratory of Future Networks of Intelligence, CUHK(SZ)  \n♡ Xi’an Jiaotong University  \nEmail: [xiaoyuma@link.cuhk.edu.cn](xiaoyuma@link.cuhk.edu.cn) , [tangxiaoying@cuhk.edu.cn](tangxiaoying@cuhk.edu.cn)  \n†Equal contribution. * Corresponding author.  \narXiv : 2608 .28067v 1 [cs .AI] 28 Aug 2026  \nAbstract  \nExisting API-only prompt optimisers are often described as interpretable, but in practice, this usually means only post-hoc inspectability:  \neach iteration still rewrites the prompt as one opaque string, leaving a trace of full-prompt diffs rather than localisable, machine-readable edits. This paper introduces SEPO (Structural, Evidence-grounded Prompt Optimization), a multi-trajectory prompt optimiser centred on edit-effect lineage feedback. Rather than treating each iteration as an isolated whole-prompt rewrite, SEPO locally edits stable, typed units in a two-layer prompt schema, links the target and realised structural operations of each edit to the examples it newly fixes or breaks, and carries this edit–effect record forward to guide later architect calls on the same search branch. This makes prompt optimisation addressable, attributable, and actionable. Across a 14-task held-out suite, SEPO improves over the strongest baseline, GEPA, by 3.1 pp on Llama-3.1-8B-Instruct and 2.2 pp on Qwen3- 8B, reaching 61.9% and 73.3% macro accuracy. SEPO also lies on both the optimisation-time and test-time Pareto frontiers, spending 2.9M optimisation tokens versus 4.1M for GEPA and producing prompts over 5 × shorter.  \n1 Introduction  \nPrompt optimisation has become a de-facto recipe for adapting frozen LLMs to new tasks without weight updates: an architect LLM proposes arewrite, the candidate is scored on a training set, and the cycle repeats (Ramnath et al., 2025 ; Yang et al., 2024 ; Pryzant et al., 2023 ; Agrawal et al., 2026 ; Yuksekgonul et al., 2025) . Despite being commonly described as interpretable, most APIonly prompt optimisers still expose the same underlying interface: each iteration rewrites the prompt  \nFigure 1: SEPO at a glance. SEPO makes localised, evidence-grounded edits (blue) to a structured prompt (top), each traceable to examples it fixed or broke, whereas monolithic rewriters rewrite the whole prompt (blue), opaquely (bottom) .  \nas one opaque string. The resulting trace is therefore a sequence of full-prompt diffs: it shows that the prompt changed, but not where the relevant change occurred, why it was made, or which examples it helped or harmed. This limitation persists whether the rewriter is driven by text-gradient feedback or by LLM-as-optimiser sampling: the unit of edit remains the entire prompt.  \nA handful of recent structured optimisers begin to challenge this single-string interface. Some expose local edit targets, such as factor-, section- , token-, or tree-level units (Schnabel and Neville, 2024 ; Liu et al., 2026c ; Sharma and Henley, 2026 ; Jain and Chowdhary, 2025 ; Kumar et al., 2025); others attribute failures to semantic units or agent contributions (Chen et al., 2026 ; Xia et al., 2026 ; Zhang et al., 2025); and reflective or self-evolving optimisers improve search with archives, critique, and synthesis (Agrawal et al., 2026 ; Agarwal et al., 2025) . Existing prompt optimisers typically update a prompt based on failures surfaced in the current  \nround. However, rules added to address those failures can also change the worker’s behaviour on examples that it handled correctly under the parent, thereby introducing new regressions. Without this parent-to-child transition, later iterations may conflate persistent failures with regressions introduced by t","cbCaiqutSrENUBw4","https://ap.wps.com/l/cbCaiqutSrENUBw4","pdf",2116159,25,"English","# Abstract\n# Introduction\n## Problem with opaque prompt diffs\n## SEPO approach: edit-effect lineage feedback\n## Evaluation setup and results","[{\"question\":\"Why do existing API-only prompt optimisers lack true interpretability?\",\"answer\":\"They usually rewrite the entire prompt as an opaque string each iteration, yielding full-prompt diff traces that do not indicate where relevant changes occurred or which examples they helped or harmed.\"},{\"question\":\"How does SEPO perform prompt optimisation differently?\",\"answer\":\"SEPO locally edits stable, typed units in a two-layer prompt schema and tracks the realised structural operations, linking each edit to the examples it newly fixes or breaks to guide later calls.\"},{\"question\":\"What performance and efficiency benefits does SEPO show?\",\"answer\":\"On a 14-task held-out suite, SEPO improves macro accuracy over the best baseline, GEPA, and reaches Pareto frontiers while using fewer optimisation tokens.\"}]","SEPO: Evidence-grounded Prompt Optimization via Structural Editing | PDF",1789426017,63]