[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-273696-105":59,"doc-detail-273696-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","task-coevolve-efficient-harness-optimization-via-adaptive-validation-task-selection","TASK-COEVOLVE - Efficient Harness Optimization via Adaptive Validation Task Selection","","Task-CoEvolve presents an efficient LLM harness optimization method based on adaptive validation task selection. Instead of updating model weights, it iteratively rewrites harness code using validation performance. Prior work evaluates candidates on a fixed full validation set each iteration, wasting computation on tasks that become non-informative as the harness changes. Task-CoEvolve co-evolves validation tasks with the harness, samples informative tasks via variance-weighted disagreement, and estimates full-set scores from partial evaluations to enable fair comparisons across iterations. Experiments on online text classification and Terminal-Bench 2.1 show consistent gains over subset baselines while reducing optimization evaluations by 80%.",{"@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":30,"@type":76,"position":81},"https://docshare.wps.com/document/technology/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/task-coevolve-efficient-harness-optimization-via-adaptive-validation-task-selection/273696/",{"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/task-coevolve-efficient-harness-optimization-via-adaptive-validation-task-selection/273696.png","ImageObject",300,407,{"name":92,"@type":93},"Sage","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-09","2026-09-15",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},"What problem does Task-CoEvolve address in harness optimization?","Question",{"text":113,"@type":114},"It targets high evaluation cost and static validation task sets in existing automated harness optimization, where every candidate is evaluated on the full fixed validation set each iteration.","Answer",{"name":116,"@type":111,"acceptedAnswer":117},"How does Task-CoEvolve choose which validation tasks to evaluate?",{"text":118,"@type":114},"It selects informative tasks where candidate harnesses disagree more, measured using variance of historical success rates, and it updates the sampling distribution as the harness evolves.",{"name":120,"@type":111,"acceptedAnswer":121},"How does Task-CoEvolve compare candidates across iterations when using different subsets?",{"text":122,"@type":114},"It estimates full-set scores from sampled tasks by using task inclusion probabilities, enabling consistent comparisons despite evaluating different subsets each iteration.","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},273696,1789472524,{"code":4,"msg":5,"data":132},{"doc_id":129,"user_id":133,"nickname":92,"user_avatar":134,"doc_module":4,"category_id":29,"category_name":30,"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":130,"read_time":145},687197207057,"https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0","TASK-COEVOLVE: EFFICIENT HARNESS OPTIMIZATION VIA ADAPTIVE VALIDATION TASK SELECTION  \nAtsuyuki Miyai Kiyoharu Aizawa∗ Toshihiko Yamasaki∗  \narXiv : 2608 .20 169v2 [cs .CL] 24 Aug 2026  \nThe University of Tokyo  \n[miyai@cvm.t.u-tokyo.ac.jp](miyai@cvm.t.u-tokyo.ac.jp)  \nExisting Harness Evolution  \nEasy Task  \nMedium Task  \nHard Task  \nFigure 1: Comparison between existing harness optimization and our Task-CoEvolve.  \nABSTRACT  \nWe present a novel approach to efficient LLM harness optimization through adaptive validation task selection. Harness optimization iteratively rewrites the harness code based on validation performance, enabling substantial performance gains without updating the underlying model weights. Existing approaches, however, evaluate a fixed validation set in full at every iteration, incurring substantial evaluation costs even on tasks that become less discriminative as the harness evolves.  \nWe propose Task-CoEvolve, which co-evolves the validation tasks with the harness by addressing two challenges: selecting informative tasks and estimating full-set performance from partial evaluations. Task-CoEvolve builds on the observation that tasks on which candidate harnesses disagree are more informative for distinguishing among them than tasks that are consistently solved or failed. It uses variance-weighted sampling based on past outcomes to focus evaluation on tasks near the capability frontier, with the sampling distribution adapting as the harness evolves. It then estimates full-set scores from the sampled tasks by accounting for their sampling probabilities, enabling consistent comparisons across iterations despite evaluating different subsets. Experiments on online text classification and Terminal-Bench 2.1 show that Task-CoEvolve consistently outperforms subset-based baselines and matches the final performance of full-set search while reducing the number of evaluations during optimization by 80% . Code will be released at [https://github.com/Agent4Science-UTokyo/](https://github.com/Agent4Science-UTokyo/)[ ](https://github.com/Agent4Science-UTokyo/)Task-CoEvolve.  \n∗ Co-supervised this work.  \n1 INTRODUCTION  \nWhen deploying LLM agents to real-world tasks, the design of the harness, the code that determines what to store, what to retrieve, and what to present to the model, is critical. Indeed, simply changing the harness around a fixed LLM has been shown to yield up to a 6 × difference in performance on the same benchmark (Tian et al., 2026), suggesting that the harness can matter as much as the underlying model itself. Traditionally, harnesses have been designed by hand (Zhang et al., 2026c ; Ye et al., 2026 ; Merrill et al., 2026 ; KRAFTON AI & Ludo Robotics, 2026), but their vast design space makes this process costly and reliant on extensive trial and error. Recently, automated harness optimization has begun to attract considerable attention, where a meta-level agent iterativelyrewrites the harness code, evaluates the resulting candidates on a benchmark, and retains changes that improve performance (Lee et al., 2026 ; Lin et al., 2026 ; Zhang et al., 2026a ; Wang et al., 2026 ; Guo et al., 2026 ; Weng, 2026) .  \nExisting work on automated harness optimization, however, has largely adopted a naive evaluation strategy: at every iteration, each candidate harness is evaluated on the entire fixed set of validation tasks (Lee et al., 2026 ; Lin et al., 2026 ; Zhang et al., 2026a ; Wang et al., 2026) . While simple to implement and straightforward for comparing candidates, this strategy has two key limitations. First, it is expensive. Every iteration requires inference over the full validation set, and when individual tasks are costly to execute, such as long-horizon terminal tasks that occupy a sandbox environment for tens of minutes (Merrill et al., 2026), evaluation can dominate the cost of the entire optimization loop. Second, it is static. As the harness evolves, the tasks that meaningfully discriminate among","cbCaifv1CSNuKeb0","https://ap.wps.com/l/cbCaifv1CSNuKeb0","pdf",490752,17,"English","# ABSTRACT\n## 1 INTRODUCTION\n## Existing Harness Evolution\n## Efficient Subset Evaluation Motivation\n## Proposed Task-CoEvolve Method","[{\"question\":\"What problem does Task-CoEvolve address in harness optimization?\",\"answer\":\"It targets high evaluation cost and static validation task sets in existing automated harness optimization, where every candidate is evaluated on the full fixed validation set each iteration.\"},{\"question\":\"How does Task-CoEvolve choose which validation tasks to evaluate?\",\"answer\":\"It selects informative tasks where candidate harnesses disagree more, measured using variance of historical success rates, and it updates the sampling distribution as the harness evolves.\"},{\"question\":\"How does Task-CoEvolve compare candidates across iterations when using different subsets?\",\"answer\":\"It estimates full-set scores from sampled tasks by using task inclusion probabilities, enabling consistent comparisons despite evaluating different subsets each iteration.\"}]","TASK-COEVOLVE - Efficient Harness Optimization via Adaptive Validation Task Selection | PDF",43]