{"id":"https://openalex.org/W7166828694","doi":"https://doi.org/10.18653/v1/2026.findings-acl.707","title":"LEPO: Latent Reasoning Policy Optimization for Large Language Models","display_name":"LEPO: Latent Reasoning Policy Optimization for Large Language Models","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166828694","doi":"https://doi.org/10.18653/v1/2026.findings-acl.707"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.707","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.707","pdf_url":"https://aclanthology.org/2026.findings-acl.707.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":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-acl.707.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139780268","display_name":"Yuyan Zhou","orcid":null},"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/A5139742262","display_name":"Jiarui Yu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiarui Yu","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/A5139718832","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/A5139800959","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/A5139845217","display_name":"Jianqing Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jianqing Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139817756","display_name":"Qiang Lin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qiang Lin","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":0.81849133,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"14416","last_page":"14427"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.21060000360012054,"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/T10028","display_name":"Topic Modeling","score":0.21060000360012054,"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.11010000109672546,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.10859999805688858,"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/language-model","display_name":"Language model","score":0.35199999809265137},{"id":"https://openalex.org/keywords/natural-language","display_name":"Natural language","score":0.2809999883174896},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.2549000084400177},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.24390000104904175},{"id":"https://openalex.org/keywords/automated-reasoning","display_name":"Automated reasoning","score":0.2433999925851822}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6237000226974487},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5493999719619751},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.35199999809265137},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3499000072479248},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.320499986410141},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.2809999883174896},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2549000084400177},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.24390000104904175},{"id":"https://openalex.org/C195344581","wikidata":"https://www.wikidata.org/wiki/Q2555318","display_name":"Automated reasoning","level":2,"score":0.2433999925851822},{"id":"https://openalex.org/C2777152325","wikidata":"https://www.wikidata.org/wiki/Q108163","display_name":"Proposition","level":2,"score":0.2409999966621399}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.707","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.707","pdf_url":"https://aclanthology.org/2026.findings-acl.707.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"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.707","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.707","pdf_url":"https://aclanthology.org/2026.findings-acl.707.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":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166828694.pdf","grobid_xml":"https://content.openalex.org/works/W7166828694.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Recently,":[0],"latent":[1,43,77,102,119],"reasoning":[2,33,44],"has":[3],"been":[4],"introduced":[5],"into":[6,42],"large":[7],"language":[8],"models":[9],"(LLMs)":[10],"to":[11,26,30,75,85],"leverage":[12],"rich":[13],"information":[14],"within":[15],"a":[16,68,96],"continuous":[17,76],"space.However,":[18],"without":[19],"stochastic":[20],"sampling,":[21,89],"these":[22],"methods":[23,115],"inevitably":[24],"collapse":[25],"deterministic":[27],"inference,":[28],"failing":[29],"discover":[31],"diverse":[32,87],"paths.To":[34],"bridge":[35],"the":[36],"gap,":[37],"we":[38,61],"inject":[39],"controllable":[40],"stochasticity":[41,84],"via":[45],"Gumbel-Softmax,":[46],"restoring":[47],"LLMs'":[48],"exploratory":[49],"capacity":[50],"and":[51,104,118],"enhancing":[52],"their":[53],"compatibility":[54],"with":[55],"Reinforcement":[56],"Learning":[57],"(RL).Building":[58],"on":[59],"this,":[60],"propose":[62],"Latent":[63],"Reasoning":[64],"Policy":[65],"Optimization":[66],"(LEPO),":[67],"novel":[69],"framework":[70],"that":[71,109],"applies":[72],"RL":[73,114],"directly":[74],"representations.Specifically,":[78],"in":[79,91],"rollout":[80],"stage,":[81,93],"LEPO":[82,94,110],"maintains":[83],"enable":[86],"trajectory":[88],"while":[90],"optimization":[92],"constructs":[95],"unified":[97],"gradient":[98],"estimation":[99],"for":[100,116],"both":[101],"representations":[103],"discrete":[105,117],"tokens.Extensive":[106],"experiments":[107],"show":[108],"significantly":[111],"outperforms":[112],"existing":[113],"reasoning.":[120]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}