{"id":"https://openalex.org/W4417092538","doi":"https://doi.org/10.18653/v1/2026.findings-acl.685","title":"DRA-GRPO: Your GRPO Needs to Know Diverse Reasoning Paths for Mathematical Reasoning","display_name":"DRA-GRPO: Your GRPO Needs to Know Diverse Reasoning Paths for Mathematical Reasoning","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W4417092538","doi":"https://doi.org/10.18653/v1/2026.findings-acl.685"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.685","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.685","pdf_url":"https://aclanthology.org/2026.findings-acl.685.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.685.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100598315","display_name":"Chen Xiwen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiwen Chen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100439574","display_name":"Wenhui Zhu","orcid":"https://orcid.org/0009-0000-5207-6283"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wenhui Zhu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5058590658","display_name":"Peijie Qiu","orcid":"https://orcid.org/0000-0002-1591-5436"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Peijie Qiu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5111345742","display_name":"Xuanzhao Dong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xuanzhao Dong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5012720408","display_name":"Hao Wang","orcid":"https://orcid.org/0000-0001-7961-7588"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hao Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5026177673","display_name":"Haiyu Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Haiyu Wu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100721217","display_name":"Huayu Li","orcid":"https://orcid.org/0009-0009-7855-3522"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huayu Li","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5002543602","display_name":"Aristeidis Sotiras","orcid":"https://orcid.org/0000-0003-0795-8820"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Aris Sotiras","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100740828","display_name":"Yalin Wang","orcid":"https://orcid.org/0000-0002-6241-735X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yalin Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5011987346","display_name":"Abolfazl Razi","orcid":"https://orcid.org/0000-0002-3330-6132"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Abolfazl Razi","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":0.02618851,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"13995","last_page":"14019"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.17149999737739563,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.17149999737739563,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T11273","display_name":"Advanced Graph Neural Networks","score":0.12790000438690186,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.08489999920129776,"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/correctness","display_name":"Correctness","score":0.7718999981880188},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.5608999729156494},{"id":"https://openalex.org/keywords/submodular-set-function","display_name":"Submodular set function","score":0.3700000047683716},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.3422999978065491},{"id":"https://openalex.org/keywords/diversity","display_name":"Diversity (politics)","score":0.34119999408721924},{"id":"https://openalex.org/keywords/bridge","display_name":"Bridge (graph theory)","score":0.33869999647140503},{"id":"https://openalex.org/keywords/situated","display_name":"Situated","score":0.32420000433921814},{"id":"https://openalex.org/keywords/semantics","display_name":"Semantics (computer science)","score":0.3160000145435333},{"id":"https://openalex.org/keywords/bridging","display_name":"Bridging (networking)","score":0.31540000438690186}],"concepts":[{"id":"https://openalex.org/C55439883","wikidata":"https://www.wikidata.org/wiki/Q360812","display_name":"Correctness","level":2,"score":0.7718999981880188},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6603000164031982},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.5608999729156494},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5015000104904175},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.38609999418258667},{"id":"https://openalex.org/C178621042","wikidata":"https://www.wikidata.org/wiki/Q7631710","display_name":"Submodular set function","level":2,"score":0.3700000047683716},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3440999984741211},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.3422999978065491},{"id":"https://openalex.org/C2781316041","wikidata":"https://www.wikidata.org/wiki/Q1230584","display_name":"Diversity (politics)","level":2,"score":0.34119999408721924},{"id":"https://openalex.org/C100776233","wikidata":"https://www.wikidata.org/wiki/Q2532492","display_name":"Bridge (graph theory)","level":2,"score":0.33869999647140503},{"id":"https://openalex.org/C132829578","wikidata":"https://www.wikidata.org/wiki/Q581151","display_name":"Situated","level":2,"score":0.32420000433921814},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.3160000145435333},{"id":"https://openalex.org/C174348530","wikidata":"https://www.wikidata.org/wiki/Q188635","display_name":"Bridging (networking)","level":2,"score":0.31540000438690186},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.31189998984336853},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.3109999895095825},{"id":"https://openalex.org/C2781311116","wikidata":"https://www.wikidata.org/wiki/Q83306","display_name":"Group (periodic table)","level":2,"score":0.2946999967098236},{"id":"https://openalex.org/C207467116","wikidata":"https://www.wikidata.org/wiki/Q4385666","display_name":"Inverse","level":2,"score":0.2915000021457672},{"id":"https://openalex.org/C152139883","wikidata":"https://www.wikidata.org/wiki/Q252973","display_name":"Mutual information","level":2,"score":0.28780001401901245},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.28760001063346863},{"id":"https://openalex.org/C195344581","wikidata":"https://www.wikidata.org/wiki/Q2555318","display_name":"Automated reasoning","level":2,"score":0.28529998660087585},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.2849000096321106},{"id":"https://openalex.org/C89611455","wikidata":"https://www.wikidata.org/wiki/Q6804646","display_name":"Mechanism (biology)","level":2,"score":0.2759999930858612},{"id":"https://openalex.org/C2778180026","wikidata":"https://www.wikidata.org/wiki/Q18378163","display_name":"Semantic heterogeneity","level":4,"score":0.2696000039577484},{"id":"https://openalex.org/C52622258","wikidata":"https://www.wikidata.org/wiki/Q131222","display_name":"Information theory","level":2,"score":0.2621000111103058},{"id":"https://openalex.org/C166052673","wikidata":"https://www.wikidata.org/wiki/Q83021","display_name":"Empirical evidence","level":2,"score":0.2605000138282776},{"id":"https://openalex.org/C149629883","wikidata":"https://www.wikidata.org/wiki/Q660926","display_name":"Fraction (chemistry)","level":2,"score":0.25529998540878296}],"mesh":[],"locations_count":5,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.685","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.685","pdf_url":"https://aclanthology.org/2026.findings-acl.685.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:2505.09655","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2505.09655","pdf_url":"https://arxiv.org/pdf/2505.09655","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":"pmh:oai:arXiv.org:2505.09655","is_oa":true,"landing_page_url":"https://arxiv.org/abs/2505.09655","pdf_url":"https://arxiv.org/pdf/2505.09655","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":"pmh:doi:10.48550/arxiv.2505.09655","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"publisher-specific-oa","license_id":"https://openalex.org/licenses/publisher-specific-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},{"id":"doi:10.48550/arxiv.2505.09655","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2505.09655","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.685","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.685","pdf_url":"https://aclanthology.org/2026.findings-acl.685.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/G1853058835","display_name":null,"funder_award_id":"CNS-2204721","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G7353134117","display_name":"CNS Core: Small: PilotPC: Proactive Inverse Learning of Network Topology for Predictive Communication among Unmanned Vehicles","funder_award_id":"2204721","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4417092538.pdf","grobid_xml":"https://content.openalex.org/works/W4417092538.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Post-training":[0],"LLMs":[1],"with":[2,31,133,156],"Reinforcement":[3],"Learning,":[4],"specifically":[5],"Group":[6],"Relative":[7],"Policy":[8],"Optimization":[9],"(GRPO),":[10],"has":[11],"emerged":[12],"as":[13],"a":[14,44,52,75,111],"paradigm":[15],"for":[16],"enhancing":[17],"mathematical":[18],"reasoning.However,":[19],"standard":[20],"GRPO":[21,134],"relies":[22],"on":[23,137,154],"scalar":[24],"correctness":[25],"rewards":[26],"that":[27,79,104,142],"are":[28],"often":[29],"non-injective":[30],"respect":[32],"to":[33,43,119],"semantic":[34,86],"content:":[35],"distinct":[36],"reasoning":[37],"paths":[38],"receive":[39],"identical":[40],"rewards.This":[41],"leads":[42],"Diversity-Quality":[45],"Inconsistency,":[46],"where":[47],"the":[48,81,85,107,117,124,165],"policy":[49,118],"collapses":[50],"into":[51],"narrow":[53],"set":[54],"of":[55,88,123,152,168],"dominant":[56],"modes":[57],"while":[58],"ignoring":[59],"equally":[60],"valid":[61],"but":[62],"structurally":[63],"novel":[64],"strategies.To":[65],"bridge":[66],"this":[67],"gap,":[68],"we":[69],"propose":[70],"Diversity-aware":[71],"Reward":[72],"Adjustment":[73],"(DRA),":[74],"theoretically":[76],"grounded":[77],"framework":[78],"calibrates":[80],"reward":[82],"signal":[83],"using":[84],"density":[87],"sampled":[89],"groups.By":[90],"leveraging":[91],"Submodular":[92],"Mutual":[93],"Information":[94],"(SMI),":[95],"DRA":[96],"implements":[97],"an":[98,149],"Inverse":[99],"Propensity":[100],"Scoring":[101],"(IPS)":[102],"mechanism":[103],"effectively":[105],"de-biases":[106],"gradient":[108],"estimation.This":[109],"creates":[110],"repulsive":[112],"force":[113],"against":[114],"redundancy,":[115],"driving":[116],"achieve":[120],"better":[121],"coverage":[122],"highreward":[125],"landscape.Our":[126],"method":[127],"is":[128,175],"plug-andplay":[129],"and":[130,161],"integrates":[131],"seamlessly":[132],"variants.Empirical":[135],"evaluations":[136],"five":[138],"math":[139],"benchmarks":[140],"demonstrate":[141],"DRA-GRPO":[143],"consistently":[144],"outperforms":[145],"strong":[146],"baselines,":[147],"achieving":[148],"average":[150],"accuracy":[151],"58.2%":[153],"DeepSeek-R1-Distill-Qwen-1.5B":[155],"only":[157],"7,000":[158],"training":[159],"samples":[160],"$55":[162],"cost,":[163],"highlighting":[164],"critical":[166],"role":[167],"diversity":[169],"calibration":[170],"in":[171],"dataefficient":[172],"alignment.The":[173],"code":[174],"available":[176],"at":[177],"https://github.com/xiwenc1/DRA-GRPO.":[178]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}