{"id":"https://openalex.org/W4416549768","doi":"https://doi.org/10.18653/v1/2026.findings-acl.875","title":"GAPO: Robust Advantage Estimation for Real-World Code LLMs","display_name":"GAPO: Robust Advantage Estimation for Real-World Code LLMs","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W4416549768","doi":"https://doi.org/10.18653/v1/2026.findings-acl.875"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.875","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.875","pdf_url":"https://aclanthology.org/2026.findings-acl.875.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":"Findings of the Association for Computational Linguistics: ACL 2026","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-acl.875.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5056834769","display_name":"Jianqing Zhang","orcid":"https://orcid.org/0009-0003-4990-8466"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jianqing Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Zhezheng Hao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhezheng Hao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5078890978","display_name":"Wei Xia","orcid":"https://orcid.org/0000-0003-2108-0807"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wei Xia","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5007017628","display_name":"Hande Dong","orcid":"https://orcid.org/0000-0003-0074-2664"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hande Dong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5110305218","display_name":"Hong Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hong Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5040060375","display_name":"Chenxing Wei","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chenxing Wei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5046553849","display_name":"Yuyan Zhou","orcid":"https://orcid.org/0000-0003-2486-5538"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yuyan Zhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5075950865","display_name":"Yanlin Qi","orcid":"https://orcid.org/0000-0001-6572-1093"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yubin Qi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102995594","display_name":"Qiang Lin","orcid":"https://orcid.org/0000-0001-9536-180X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qiang Lin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5112582790","display_name":"Jian Cao","orcid":"https://orcid.org/0009-0003-2110-2620"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jian Cao","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":0.01268387,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"17663","last_page":"17673"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10260","display_name":"Software Engineering Research","score":0.41200000047683716,"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"}},"topics":[{"id":"https://openalex.org/T10260","display_name":"Software Engineering Research","score":0.41200000047683716,"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/T10743","display_name":"Software Testing and Debugging Techniques","score":0.09229999780654907,"subfield":{"id":"https://openalex.org/subfields/1712","display_name":"Software"},"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/T10181","display_name":"Natural Language Processing Techniques","score":0.07980000227689743,"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/code","display_name":"Code (set theory)","score":0.6079000234603882},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.5993000268936157},{"id":"https://openalex.org/keywords/clipping","display_name":"Clipping (morphology)","score":0.5637999773025513},{"id":"https://openalex.org/keywords/interval","display_name":"Interval (graph theory)","score":0.5580000281333923},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.47600001096725464},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.38769999146461487},{"id":"https://openalex.org/keywords/matching","display_name":"Matching (statistics)","score":0.3321000039577484}],"concepts":[{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.6079000234603882},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.5993000268936157},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5968000292778015},{"id":"https://openalex.org/C2776848632","wikidata":"https://www.wikidata.org/wiki/Q853463","display_name":"Clipping (morphology)","level":2,"score":0.5637999773025513},{"id":"https://openalex.org/C2778067643","wikidata":"https://www.wikidata.org/wiki/Q166507","display_name":"Interval (graph theory)","level":2,"score":0.5580000281333923},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.47600001096725464},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4099999964237213},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.38769999146461487},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.35199999809265137},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.3321000039577484},{"id":"https://openalex.org/C96250715","wikidata":"https://www.wikidata.org/wiki/Q965330","display_name":"Estimation","level":2,"score":0.3296999931335449},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.32350000739097595},{"id":"https://openalex.org/C205167067","wikidata":"https://www.wikidata.org/wiki/Q3300636","display_name":"Interval estimation","level":3,"score":0.31709998846054077},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.314300000667572},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3093999922275543},{"id":"https://openalex.org/C115051666","wikidata":"https://www.wikidata.org/wiki/Q6522493","display_name":"Ranging","level":2,"score":0.29440000653266907},{"id":"https://openalex.org/C37404715","wikidata":"https://www.wikidata.org/wiki/Q380679","display_name":"Dynamic programming","level":2,"score":0.2775999903678894},{"id":"https://openalex.org/C103402496","wikidata":"https://www.wikidata.org/wiki/Q1106171","display_name":"Prediction interval","level":2,"score":0.2651999890804291},{"id":"https://openalex.org/C191252586","wikidata":"https://www.wikidata.org/wiki/Q1671453","display_name":"Interval arithmetic","level":3,"score":0.25529998540878296}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.875","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.875","pdf_url":"https://aclanthology.org/2026.findings-acl.875.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":"Findings of the Association for Computational Linguistics: ACL 2026","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2510.21830","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.21830","pdf_url":"https://arxiv.org/pdf/2510.21830","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,"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.2510.21830","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2510.21830","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,"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":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.875","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.875","pdf_url":"https://aclanthology.org/2026.findings-acl.875.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":"Findings of the Association for Computational Linguistics: ACL 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G5847459458","display_name":null,"funder_award_id":"YG2024QNB05","funder_id":"https://openalex.org/F4320322999","funder_display_name":"Shanghai Jiao Tong University"},{"id":"https://openalex.org/G6188036497","display_name":null,"funder_award_id":"2025ZHYL003","funder_id":"https://openalex.org/F4320317789","funder_display_name":"Shanghai Municipal Health Commission"}],"funders":[{"id":"https://openalex.org/F4320316083","display_name":"Tencent","ror":"https://ror.org/00hhjss72"},{"id":"https://openalex.org/F4320317789","display_name":"Shanghai Municipal Health Commission","ror":null},{"id":"https://openalex.org/F4320322999","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04"},{"id":"https://openalex.org/F4320335227","display_name":"Bijzonder Onderzoeksfonds UGent","ror":"https://ror.org/00cv9y106"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416549768.pdf","grobid_xml":"https://content.openalex.org/works/W4416549768.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Reinforcement":[0],"learning":[1],"(RL)":[2],"is":[3],"widely":[4],"used":[5],"for":[6],"post-training":[7],"large":[8,119],"language":[9],"models":[10],"(LLMs)":[11],"in":[12,31,91],"code":[13],"editing,":[14],"where":[15],"group-relative":[16],"methods,":[17],"such":[18],"as":[19,82],"GRPO,":[20],"are":[21,37],"popular":[22],"due":[23],"to":[24,44,86,94,133],"their":[25],"critic-free":[26],"and":[27,48,75,107,137,145,154],"normalized":[28],"advantage":[29,46,92],"estimation.However,":[30],"real-world":[32],"code-editing":[33,125],"scenarios,":[34],"reward":[35],"distributions":[36],"often":[38],"skewed":[39],"with":[40,67],"unpredictable":[41],"noise,":[42],"leading":[43],"distorted":[45],"computation":[47],"increased":[49],"rollout":[50,102],"outliers.To":[51],"address":[52],"this":[53],"issue,":[54],"we":[55],"propose":[56],"Group":[57],"Adaptive":[58],"Policy":[59],"Optimization":[60],"(GAPO),":[61],"which":[62],"adaptively":[63],"finds":[64],"an":[65,83],"interval":[66,81],"the":[68,77,88],"highest":[69],"SNR":[70],"(signal-to-noise":[71],"Ratio)":[72],"per":[73],"prompt":[74],"uses":[76],"median":[78],"of":[79,121],"that":[80],"adaptive":[84,98],"Q":[85,99],"replace":[87],"group":[89],"mean":[90],"calculation":[93],"reduce":[95],"noise":[96,103],"further.This":[97],"robustly":[100],"handles":[101],"while":[104,149],"remaining":[105],"plug-and-play":[106],"efficient.We":[108],"evaluate":[109],"GAPO":[110],"on":[111],"nine":[112],"instruction-tuned":[113],"LLMs":[114],"(3B-14B)":[115],"using":[116],"a":[117],"collected":[118],"dataset":[120],"51,844":[122],"real-world,":[123],"history-aware":[124],"tasks":[126],"spanning":[127],"10":[128],"programming":[129],"languages.GAPO":[130],"yields":[131],"up":[132],"4.35":[134],"in-domain":[135],"(ID)":[136],"5.30":[138],"out-of-domain":[139],"(OOD)":[140],"exact-match":[141],"improvements":[142],"over":[143],"GRPO":[144],"its":[146],"variant":[147],"DAPO,":[148],"achieving":[150],"lower":[151],"clipping":[152],"ratios":[153],"higher":[155],"GPU":[156],"throughput.":[157]},"counts_by_year":[],"updated_date":"2026-07-31T08:31:51.225901","created_date":"2025-10-29T00:00:00"}