{"id":"https://openalex.org/W4416529508","doi":"https://doi.org/10.52202/085713-2259","title":"Right Question is Already Half the Answer: Fully Unsupervised LLM Reasoning Incentivization","display_name":"Right Question is Already Half the Answer: Fully Unsupervised LLM Reasoning Incentivization","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416529508","doi":"https://doi.org/10.52202/085713-2259"},"language":"en","primary_location":{"id":"doi:10.52202/085713-2259","doi":"https://doi.org/10.52202/085713-2259","is_oa":false,"landing_page_url":"https://doi.org/10.52202/085713-2259","pdf_url":null,"source":{"id":"https://openalex.org/S4363606243","display_name":"neural information processing systems","issn_l":"1049-5258","issn":["1049-5258"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","norway-2"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Advances in Neural Information Processing Systems 38","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2504.05812","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5120704486","display_name":"Qingyang Zhang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I162868743","display_name":"Tianjin University","ror":"https://ror.org/012tb2g32","country_code":"CN","type":"education","lineage":["https://openalex.org/I162868743"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Qingyang Zhang","raw_affiliation_strings":["Tianjin University  Tencent AI Lab"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tianjin University  Tencent AI Lab","institution_ids":["https://openalex.org/I162868743"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100757788","display_name":"Haitao Wu","orcid":"https://orcid.org/0000-0001-5164-5501","observed_orcids":["https://orcid.org/0000-0001-5164-5501"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Haitao Wu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Changqing Zhang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I162868743","display_name":"Tianjin University","ror":"https://ror.org/012tb2g32","country_code":"CN","type":"education","lineage":["https://openalex.org/I162868743"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Changqing Zhang","raw_affiliation_strings":["Tianjin University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tianjin University","institution_ids":["https://openalex.org/I162868743"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Peilin Zhao","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Peilin Zhao","raw_affiliation_strings":["Tencent"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tencent","institution_ids":["https://openalex.org/I2250653659"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5045777220","display_name":"Yatao Bian","orcid":"https://orcid.org/0000-0002-2368-4084","observed_orcids":["https://orcid.org/0000-0002-2368-4084"]},"institutions":[{"id":"https://openalex.org/I165932596","display_name":"National University of Singapore","ror":"https://ror.org/01tgyzw49","country_code":"SG","type":"education","lineage":["https://openalex.org/I165932596"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Yatao Bian","raw_affiliation_strings":["National University of Singapore"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"National University of Singapore","institution_ids":["https://openalex.org/I165932596"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":1.8965,"has_fulltext":true,"cited_by_count":2,"citation_normalized_percentile":{"value":0.8795466,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":96,"max":97},"biblio":{"volume":null,"issue":null,"first_page":"75038","last_page":"75065"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.8227999806404114,"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":0.8227999806404114,"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":0.06830000132322311,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.02290000021457672,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/large-language-model-reasoning","display_name":"large language model reasoning","score":0.8220000267028809},{"id":"https://openalex.org/keywords/entropy-minimization","display_name":"entropy minimization","score":0.6579999923706055},{"id":"https://openalex.org/keywords/predictive-entropy","display_name":"predictive entropy","score":0.6959999799728394},{"id":"https://openalex.org/keywords/latent-semantic-space","display_name":"latent semantic space","score":0.7039999961853027},{"id":"https://openalex.org/keywords/mathematical-reasoning","display_name":"mathematical reasoning","score":0.7820000052452087},{"id":"https://openalex.org/keywords/natural-language-reasoning","display_name":"natural language reasoning","score":0.6320000290870667},{"id":"https://openalex.org/keywords/mmlu-pro","display_name":"MMLU-Pro","score":0.7360000014305115},{"id":"https://openalex.org/keywords/qwen2-5","display_name":"Qwen2.5","score":0.828000009059906}],"concepts":[{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.7228000164031982},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6265000104904175},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5856000185012817},{"id":"https://openalex.org/C106301342","wikidata":"https://www.wikidata.org/wiki/Q4117933","display_name":"Entropy (arrow of time)","level":2,"score":0.5378999710083008},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.49410000443458557},{"id":"https://openalex.org/C8038995","wikidata":"https://www.wikidata.org/wiki/Q1152135","display_name":"Unsupervised learning","level":2,"score":0.484499990940094},{"id":"https://openalex.org/C4554734","wikidata":"https://www.wikidata.org/wiki/Q593744","display_name":"Knowledge base","level":2,"score":0.42719998955726624},{"id":"https://openalex.org/C136389625","wikidata":"https://www.wikidata.org/wiki/Q334384","display_name":"Supervised learning","level":3,"score":0.42590001225471497},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.3831000030040741},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.34860000014305115},{"id":"https://openalex.org/C58973888","wikidata":"https://www.wikidata.org/wiki/Q1041418","display_name":"Semi-supervised learning","level":2,"score":0.29660001397132874},{"id":"https://openalex.org/C44291984","wikidata":"https://www.wikidata.org/wiki/Q1074173","display_name":"Question answering","level":2,"score":0.28839999437332153},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.2770000100135803},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.2761000096797943},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.26589998602867126},{"id":"https://openalex.org/C9679016","wikidata":"https://www.wikidata.org/wiki/Q1417473","display_name":"Principle of maximum entropy","level":2,"score":0.2655999958515167}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.52202/085713-2259","doi":"https://doi.org/10.52202/085713-2259","is_oa":false,"landing_page_url":"https://doi.org/10.52202/085713-2259","pdf_url":null,"source":{"id":"https://openalex.org/S4363606243","display_name":"neural information processing systems","issn_l":"1049-5258","issn":["1049-5258"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","norway-2"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Advances in Neural Information Processing Systems 38","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2504.05812","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2504.05812","pdf_url":"https://arxiv.org/pdf/2504.05812","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2504.05812","doi":"https://doi.org/10.48550/arxiv.2504.05812","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2504.05812","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2504.05812","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2504.05812","pdf_url":"https://arxiv.org/pdf/2504.05812","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","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/W4416529508.pdf","grobid_xml":"https://content.openalex.org/works/W4416529508.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7160561377","https://openalex.org/W4414829284","https://openalex.org/W4416600559","https://openalex.org/W4404396425","https://openalex.org/W4416045951","https://openalex.org/W4415328503","https://openalex.org/W4414281281","https://openalex.org/W7127636261","https://openalex.org/W4416131239","https://openalex.org/W7139028173"],"abstract_inverted_index":{"Existing":[0],"methods":[1],"to":[2,85,110,124,135],"enhance":[3],"the":[4,67,103,117,137],"reasoning":[5,34,62,94],"capability":[6],"of":[7,70,105,119,139],"large":[8],"language":[9],"models":[10],"predominantly":[11],"rely":[12],"on":[13,22,29,72,88,112,126],"supervised":[14,86,99],"fine-tuning":[15],"(SFT)":[16],"followed":[17],"by":[18],"reinforcement":[19],"learning":[20],"(RL)":[21],"reasoning-specific":[23],"data.":[24],"These":[25],"approaches":[26],"critically":[27],"depend":[28],"external":[30],"supervisions--such":[31],"as":[32],"labeled":[33],"traces,":[35],"verified":[36],"golden":[37],"answers,":[38],"or":[39],"pre-trained":[40],"reward":[41],"models.":[42],"In":[43],"this":[44],"work,":[45],"we":[46],"propose":[47],"Entropy":[48],"Minimized":[49],"Policy":[50],"Optimization":[51],"(\\ours),":[52],"which":[53],"makes":[54],"an":[55],"early":[56],"attempt":[57],"at":[58,144],"fully":[59],"unsupervised":[60],"LLM":[61],"incentivization.":[63],"By":[64],"continuously":[65],"minimizing":[66],"predictive":[68],"entropy":[69],"LLMs":[71],"unlabeled":[73],"questions":[74],"in":[75],"a":[76],"latent":[77],"semantic":[78],"space,":[79],"\\ours":[80,101],"achieves":[81],"competitive":[82],"performance":[83],"compared":[84],"counterparts":[87],"both":[89],"mathematical":[90,113],"and":[91,115,130],"free-form":[92],"natural":[93],"tasks.":[95],"Specifically,":[96],"without":[97],"any":[98],"signals,":[100],"boosts":[102],"accuracy":[104,118],"Qwen2.5-Math-7B":[106],"Base":[107,121],"from":[108,122],"30.7\\%":[109],"48.1\\%":[111],"benchmarks":[114],"improves":[116],"Qwen2.5-7B":[120],"32.1\\%":[123],"50.1\\%":[125],"MMLU-Pro.":[127],"Primary":[128],"experiments":[129],"analysis":[131],"are":[132],"also":[133],"provided":[134],"interpret":[136],"effectiveness":[138],"\\ours.":[140],"Code":[141],"is":[142],"available":[143],"https://github.com/QingyangZhang/EMPO.":[145]},"counts_by_year":[{"year":2026,"cited_by_count":2}],"updated_date":"2026-10-06T10:28:23.643775","created_date":"2025-10-10T00:00:00"}