{"id":"https://openalex.org/W7166842306","doi":"https://doi.org/10.18653/v1/2026.acl-long.504","title":"Learning from the Irrecoverable: Error-Localized Policy Optimization for Tool-Integrated LLM Reasoning","display_name":"Learning from the Irrecoverable: Error-Localized Policy Optimization for Tool-Integrated LLM Reasoning","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166842306","doi":"https://doi.org/10.18653/v1/2026.acl-long.504"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.acl-long.504","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.504","pdf_url":"https://aclanthology.org/2026.acl-long.504.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.acl-long.504.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5048547376","display_name":"Liang Qiao","orcid":"https://orcid.org/0000-0002-6778-4431"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qiao Liang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030826237","display_name":"Yuke Zhu","orcid":"https://orcid.org/0000-0002-9198-2227"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yuke Zhu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139783425","display_name":"Chao Ge","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chao Ge","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139786556","display_name":"Lei Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lei Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139779280","display_name":"Ying Shen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ying Shen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139732377","display_name":"Bo Zheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bo Zheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5100752605","display_name":"Sheng Guo","orcid":"https://orcid.org/0000-0002-8959-2473"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sheng Guo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.83224279,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"11008","last_page":"11028"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11010","display_name":"Logic, Reasoning, and Knowledge","score":0.2612000107765198,"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/T11010","display_name":"Logic, Reasoning, and Knowledge","score":0.2612000107765198,"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/T10456","display_name":"Multi-Agent Systems and Negotiation","score":0.2290000021457672,"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/T10927","display_name":"Access Control and Trust","score":0.12150000035762787,"subfield":{"id":"https://openalex.org/subfields/3312","display_name":"Sociology and Political Science"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.3037000000476837},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.2874999940395355},{"id":"https://openalex.org/keywords/case-based-reasoning","display_name":"Case-based reasoning","score":0.2842999994754791},{"id":"https://openalex.org/keywords/matching","display_name":"Matching (statistics)","score":0.274399995803833},{"id":"https://openalex.org/keywords/action","display_name":"Action (physics)","score":0.2680000066757202}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5677000284194946},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.44029998779296875},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.3037000000476837},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2874999940395355},{"id":"https://openalex.org/C539667460","wikidata":"https://www.wikidata.org/wiki/Q2414942","display_name":"Management science","level":1,"score":0.28619998693466187},{"id":"https://openalex.org/C20162079","wikidata":"https://www.wikidata.org/wiki/Q1151406","display_name":"Case-based reasoning","level":2,"score":0.2842999994754791},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.274399995803833},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.2680000066757202},{"id":"https://openalex.org/C9652623","wikidata":"https://www.wikidata.org/wiki/Q190109","display_name":"Field (mathematics)","level":2,"score":0.2655999958515167},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.2590999901294708}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.acl-long.504","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.504","pdf_url":"https://aclanthology.org/2026.acl-long.504.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.acl-long.504","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.504","pdf_url":"https://aclanthology.org/2026.acl-long.504.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G5825415874","display_name":null,"funder_award_id":"62476202","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320329860","display_name":"National Science and Technology Major Project","ror":null},{"id":"https://openalex.org/F4320335787","display_name":"Fundamental Research Funds for the Central Universities","ror":null}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166842306.pdf","grobid_xml":"https://content.openalex.org/works/W7166842306.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Tool-integrated":[0],"reasoning":[1],"(TIR)":[2],"enables":[3],"LLM":[4],"agents":[5],"to":[6,48,98],"solve":[7],"tasks":[8],"through":[9,89],"planning,":[10],"tool":[11],"use,":[12],"and":[13,28,54,93,106,115,134,140],"iterative":[14],"revision,":[15],"but":[16],"outcome-only":[17],"reinforcement":[18],"learning":[19,87],"in":[20,111,132],"this":[21],"setting":[22],"suffers":[23],"from":[24],"sparse,":[25],"delayed":[26],"rewards":[27],"weak":[29],"step-level":[30],"credit":[31,59],"assignment.In":[32],"long-horizon":[33],"TIR":[34,109],"trajectories,":[35],"an":[36],"early":[37],"irrecoverable":[38,52,70],"mistake":[39],"can":[40],"determine":[41],"success":[42],"or":[43],"failure,":[44],"making":[45],"it":[46,56],"crucial":[47],"localize":[49],"the":[50,68,82,103],"first":[51,69],"step":[53,71,105],"leverage":[55],"for":[57],"fine-grained":[58],"assignment.We":[60],"propose":[61],"Error-Localized":[62],"Policy":[63],"Optimization":[64],"(ELPO),":[65],"which":[66],"localizes":[67],"via":[72],"binary-search":[73],"rollout":[74,79,137],"trees":[75],"under":[76,125],"a":[77],"fixed":[78],"budget,":[80],"converts":[81],"resulting":[83],"tree":[84],"into":[85],"stable":[86],"signals":[88],"hierarchical":[90],"advantage":[91],"attribution,":[92],"applies":[94],"error-localized":[95],"adaptive":[96],"clipping":[97],"strengthen":[99],"corrective":[100],"updates":[101],"on":[102],"critical":[104],"its":[107],"suffix.Across":[108],"benchmarks":[110],"math,":[112],"science":[113],"QA,":[114],"code":[116],"execution,":[117],"ELPO":[118],"consistently":[119],"outperforms":[120],"strong":[121],"Agentic":[122],"RL":[123],"baselines":[124],"comparable":[126],"sampling":[127],"budgets,":[128],"with":[129],"additional":[130],"gains":[131],"Pass@K":[133],"Major@K":[135],"scaling,":[136],"ranking":[138],"quality,":[139],"tool-call":[141],"efficiency.":[142]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}