{"id":"https://openalex.org/W7214890487","doi":"https://doi.org/10.48550/arxiv.2609.37968","title":"SelfSearch: Reward-Free Search for Self-Improving Agents","display_name":"SelfSearch: Reward-Free Search for Self-Improving Agents","publication_year":2026,"publication_date":"2026-09-29","ids":{"openalex":"https://openalex.org/W7214890487","doi":"https://doi.org/10.48550/arxiv.2609.37968"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.37968","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.37968","pdf_url":"https://arxiv.org/pdf/2609.37968","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2609.37968","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5153664264","display_name":"Jungwoo Yang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Yang, Jungwoo","raw_affiliation_strings":["Graduate School of Data Science , Seoul National University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Graduate School of Data Science , Seoul National University","institution_ids":["https://openalex.org/I139264467"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5153565690","display_name":"In Jin Kong","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Kong, In Jin","raw_affiliation_strings":["Graduate School of Data Science , Seoul National University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Graduate School of Data Science , Seoul National University","institution_ids":["https://openalex.org/I139264467"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5153561830","display_name":"Yohan Jo","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Jo, Yohan","raw_affiliation_strings":["Graduate School of Data Science , Seoul National University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Graduate School of Data Science , Seoul National University","institution_ids":["https://openalex.org/I139264467"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I139264467"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":2,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11975","display_name":"Evolutionary Algorithms and Applications","score":0.4828999936580658,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11975","display_name":"Evolutionary Algorithms and Applications","score":0.4828999936580658,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10260","display_name":"Software Engineering Research","score":0.2020999938249588,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10906","display_name":"AI-based Problem Solving and Planning","score":0.17260000109672546,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/self-improving-agents","display_name":"self-improving agents","score":0.906000018119812},{"id":"https://openalex.org/keywords/self-modification","display_name":"self-modification","score":0.765999972820282},{"id":"https://openalex.org/keywords/llm-based-agents","display_name":"LLM agents","score":0.8349999785423279},{"id":"https://openalex.org/keywords/experiential-learning","display_name":"experiential learning","score":0.36000001430511475},{"id":"https://openalex.org/keywords/search-cost-reduction","display_name":"search cost reduction","score":0.6190000176429749},{"id":"https://openalex.org/keywords/swe-bench","display_name":"SWE-bench","score":0.8740000128746033},{"id":"https://openalex.org/keywords/code-generation","display_name":"code generation","score":0.34299999475479126}],"concepts":[{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.6783999800682068},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.647599995136261},{"id":"https://openalex.org/C2776207758","wikidata":"https://www.wikidata.org/wiki/Q5303302","display_name":"Downstream (manufacturing)","level":2,"score":0.6316999793052673},{"id":"https://openalex.org/C179518139","wikidata":"https://www.wikidata.org/wiki/Q5140297","display_name":"Coding (social sciences)","level":2,"score":0.4650999903678894},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.46000000834465027},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3700000047683716},{"id":"https://openalex.org/C19889080","wikidata":"https://www.wikidata.org/wiki/Q2835852","display_name":"Beam search","level":3,"score":0.3546000123023987},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.35249999165534973},{"id":"https://openalex.org/C21782646","wikidata":"https://www.wikidata.org/wiki/Q841666","display_name":"Search cost","level":2,"score":0.3086000084877014},{"id":"https://openalex.org/C74072328","wikidata":"https://www.wikidata.org/wiki/Q1142726","display_name":"Intelligent agent","level":2,"score":0.2922999858856201},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2897000014781952},{"id":"https://openalex.org/C2775935494","wikidata":"https://www.wikidata.org/wiki/Q741964","display_name":"Search and rescue","level":3,"score":0.2639000117778778},{"id":"https://openalex.org/C97854310","wikidata":"https://www.wikidata.org/wiki/Q19541","display_name":"Search engine","level":2,"score":0.258899986743927},{"id":"https://openalex.org/C125583679","wikidata":"https://www.wikidata.org/wiki/Q755673","display_name":"Search algorithm","level":2,"score":0.2531999945640564}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.37968","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.37968","pdf_url":"https://arxiv.org/pdf/2609.37968","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2609.37968","doi":"https://doi.org/10.48550/arxiv.2609.37968","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.37968","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2609.37968","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.37968","pdf_url":"https://arxiv.org/pdf/2609.37968","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7214890487.pdf","grobid_xml":"https://content.openalex.org/works/W7214890487.grobid-xml"},"referenced_works_count":2,"referenced_works":["https://openalex.org/W7127903492","https://openalex.org/W7140288800"],"related_works":["https://openalex.org/W4412889810","https://openalex.org/W7160947383","https://openalex.org/W7202208313","https://openalex.org/W6948264221","https://openalex.org/W7202196053","https://openalex.org/W7125367450","https://openalex.org/W4403963915","https://openalex.org/W4416057189","https://openalex.org/W7172337529","https://openalex.org/W7108325177"],"abstract_inverted_index":{"Advances":[0],"in":[1,54,102,162],"the":[2,41,44,68,99,141,181,189],"coding":[3],"capabilities":[4,205],"of":[5,61,74,172,183],"LLM":[6],"agents":[7,30,56,109],"allow":[8],"them":[9],"to":[10,26,43,112],"inspect":[11],"and":[12,18,39,72,86,143,206],"modify":[13,57],"their":[14],"own":[15],"instructions,":[16],"tools,":[17],"execution":[19,132],"procedures.":[20],"Existing":[21],"approaches":[22],"use":[23],"this":[24],"ability":[25],"search":[27,42,52,153,157,163],"for":[28,81],"improved":[29],"through":[31,199],"repeated":[32],"downstream":[33,89,204],"evaluation,":[34],"which":[35,55],"incurs":[36],"substantial":[37],"costs":[38],"ties":[40],"evaluated":[45],"tasks.":[46],"We":[47],"introduce":[48],"\\textbf{SelfSearch},":[49],"a":[50,167,184],"reward-free":[51],"procedure":[53],"themselves":[58],"using":[59],"records":[60,66],"previous":[62],"self-improvement":[63],"episodes.":[64],"These":[65,193],"capture":[67],"reasoning,":[69],"tool":[70],"actions,":[71],"outcomes":[73],"earlier":[75],"modification":[76],"attempts,":[77],"providing":[78],"concrete":[79],"experience":[80,197],"improving":[82],"both":[83,140],"task":[84,149],"solving":[85],"self-modification.":[87],"Without":[88],"reward":[90],"signals":[91],"during":[92],"search,":[93],"SelfSearch":[94,146],"improves":[95,124],"population-mean":[96],"success":[97,125,150],"over":[98],"initial":[100,142],"agent":[101,123],"all":[103],"six":[104],"model--benchmark":[105],"settings,":[106],"with":[107,151,176],"individual":[108],"gaining":[110],"up":[111],"11.2":[113],"percentage":[114,128],"points":[115,129],"on":[116,136],"Terminal-Bench":[117,173],"2.1.":[118],"On":[119],"SWE-bench":[120],"Multilingual,":[121],"an":[122],"by":[126,134,139],"\\textbf{5.0}":[127],"while":[130],"reducing":[131],"cost":[133],"\\textbf{38.5}\\%":[135],"tasks":[137,175],"solved":[138],"evolved":[144],"agents.":[145],"achieves":[147],"competitive":[148],"evaluation-guided":[152],"baselines":[154],"at":[155],"lower":[156],"cost.":[158],"With":[159],"only":[160],"\\textbf{\\$4.03}":[161],"cost,":[164],"it":[165],"produces":[166],"harness":[168],"that":[169,196],"solves":[170],"\\textbf{82.0}\\%":[171],"2.1":[174],"DeepSeek":[177],"V4":[178],"Flash":[179],"under":[180],"settings":[182],"public":[185],"nine-harness":[186],"comparison,":[187],"matching":[188],"top-scoring":[190],"harness,":[191],"Codex.":[192],"results":[194],"suggest":[195],"gained":[198],"self-modification":[200],"can":[201],"improve":[202],"agents'":[203],"efficiency.":[207]},"counts_by_year":[{"year":2026,"cited_by_count":2}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-10-01T00:00:00"}