{"id":"https://openalex.org/W4415194397","doi":"https://doi.org/10.48550/arxiv.2505.19590","title":"Learning to Reason without External Rewards","display_name":"Learning to Reason without External Rewards","publication_year":2025,"publication_date":"2025-05-26","ids":{"openalex":"https://openalex.org/W4415194397","doi":"https://doi.org/10.48550/arxiv.2505.19590"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2505.19590","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2505.19590","pdf_url":"https://arxiv.org/pdf/2505.19590","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2505.19590","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5120003730","display_name":"Xuandong Zhao","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I95457486","display_name":"University of California, Berkeley","ror":"https://ror.org/01an7q238","country_code":"US","type":"education","lineage":["https://openalex.org/I95457486"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Zhao, Xuandong","raw_affiliation_strings":["UC Berkeley"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"UC Berkeley","institution_ids":["https://openalex.org/I95457486"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120003731","display_name":"Zhewei Kang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I95457486","display_name":"University of California, Berkeley","ror":"https://ror.org/01an7q238","country_code":"US","type":"education","lineage":["https://openalex.org/I95457486"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Kang, Zhewei","raw_affiliation_strings":["UC Berkeley"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"UC Berkeley","institution_ids":["https://openalex.org/I95457486"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5109022095","display_name":"Aosong Feng","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I32971472","display_name":"Yale University","ror":"https://ror.org/03v76x132","country_code":"US","type":"education","lineage":["https://openalex.org/I32971472"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Feng, Aosong","raw_affiliation_strings":["Yale University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Yale University","institution_ids":["https://openalex.org/I32971472"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5026322200","display_name":"Sergey Levine","orcid":"https://orcid.org/0000-0001-6764-2743","observed_orcids":["https://orcid.org/0000-0001-6764-2743"]},"institutions":[{"id":"https://openalex.org/I95457486","display_name":"University of California, Berkeley","ror":"https://ror.org/01an7q238","country_code":"US","type":"education","lineage":["https://openalex.org/I95457486"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Levine, Sergey","raw_affiliation_strings":["UC Berkeley"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"UC Berkeley","institution_ids":["https://openalex.org/I95457486"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5019426968","display_name":"Dawn Xiaodong Song","orcid":"https://orcid.org/0000-0001-9745-6802","observed_orcids":["https://orcid.org/0000-0001-9745-6802"]},"institutions":[{"id":"https://openalex.org/I95457486","display_name":"University of California, Berkeley","ror":"https://ror.org/01an7q238","country_code":"US","type":"education","lineage":["https://openalex.org/I95457486"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Song, Dawn","raw_affiliation_strings":["UC Berkeley"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"UC Berkeley","institution_ids":["https://openalex.org/I95457486"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T13308","display_name":"Cybernetics and Technology in Society","score":0.3813000023365021,"subfield":{"id":"https://openalex.org/subfields/1207","display_name":"History and Philosophy of Science"},"field":{"id":"https://openalex.org/fields/12","display_name":"Arts and Humanities"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},"topics":[{"id":"https://openalex.org/T13308","display_name":"Cybernetics and Technology in Society","score":0.3813000023365021,"subfield":{"id":"https://openalex.org/subfields/1207","display_name":"History and Philosophy of Science"},"field":{"id":"https://openalex.org/fields/12","display_name":"Arts and Humanities"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T10258","display_name":"Philosophy and Theoretical Science","score":0.195700004696846,"subfield":{"id":"https://openalex.org/subfields/3205","display_name":"Experimental and Cognitive Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T14381","display_name":"Psychiatry, Mental Health, Neuroscience","score":0.08829999715089798,"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/ai-alignment","display_name":"AI alignment","score":0.3269999921321869},{"id":"https://openalex.org/keywords/self-image","display_name":"self-image","score":0.47999998927116394},{"id":"https://openalex.org/keywords/ai-emotions","display_name":"AI emotions","score":0.4690000116825104},{"id":"https://openalex.org/keywords/ai-self-awareness","display_name":"AI self-awareness","score":0.3050000071525574}],"concepts":[{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.7542999982833862},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.7416999936103821},{"id":"https://openalex.org/C85847156","wikidata":"https://www.wikidata.org/wiki/Q59015987","display_name":"Verifiable secret sharing","level":3,"score":0.6492999792098999},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6244000196456909},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.573199987411499},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.5432000160217285},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.5110999941825867},{"id":"https://openalex.org/C67203356","wikidata":"https://www.wikidata.org/wiki/Q1321905","display_name":"Reinforcement","level":2,"score":0.4343000054359436},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4113999903202057},{"id":"https://openalex.org/C77967617","wikidata":"https://www.wikidata.org/wiki/Q4677561","display_name":"Active learning (machine learning)","level":2,"score":0.31029999256134033},{"id":"https://openalex.org/C2777267654","wikidata":"https://www.wikidata.org/wiki/Q3519023","display_name":"Test (biology)","level":2,"score":0.30329999327659607},{"id":"https://openalex.org/C12298181","wikidata":"https://www.wikidata.org/wiki/Q7246814","display_name":"Proactive learning","level":5,"score":0.2854999899864197},{"id":"https://openalex.org/C12713177","wikidata":"https://www.wikidata.org/wiki/Q1900281","display_name":"Perspective (graphical)","level":2,"score":0.2759000062942505},{"id":"https://openalex.org/C2778069772","wikidata":"https://www.wikidata.org/wiki/Q2119819","display_name":"Code of conduct","level":2,"score":0.275299996137619},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.2639999985694885},{"id":"https://openalex.org/C180747234","wikidata":"https://www.wikidata.org/wiki/Q23373","display_name":"Cognitive psychology","level":1,"score":0.2558000087738037},{"id":"https://openalex.org/C8038995","wikidata":"https://www.wikidata.org/wiki/Q1152135","display_name":"Unsupervised learning","level":2,"score":0.25529998540878296},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.25360000133514404},{"id":"https://openalex.org/C132758656","wikidata":"https://www.wikidata.org/wiki/Q5307365","display_name":"Dreyfus model of skill acquisition","level":2,"score":0.25049999356269836}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2505.19590","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2505.19590","pdf_url":"https://arxiv.org/pdf/2505.19590","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2505.19590","doi":"https://doi.org/10.48550/arxiv.2505.19590","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2505.19590","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2505.19590","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2505.19590","pdf_url":"https://arxiv.org/pdf/2505.19590","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":null,"license_id":null,"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/W4415194397.pdf","grobid_xml":"https://content.openalex.org/works/W4415194397.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7206200287","https://openalex.org/W7204840993","https://openalex.org/W7204941089","https://openalex.org/W7206161151","https://openalex.org/W7204628046","https://openalex.org/W7204221633","https://openalex.org/W7204766493","https://openalex.org/W7206159155","https://openalex.org/W7211888802","https://openalex.org/W7212935872"],"abstract_inverted_index":{"Training":[0],"large":[1],"language":[2],"models":[3],"(LLMs)":[4],"for":[5,129],"complex":[6],"reasoning":[7],"via":[8],"Reinforcement":[9,27],"Learning":[10,28],"with":[11,76],"Verifiable":[12],"Rewards":[13],"(RLVR)":[14],"is":[15,139],"effective":[16,119],"but":[17],"limited":[18],"by":[19],"reliance":[20],"on":[21,90],"costly,":[22],"domain-specific":[23],"supervision.":[24],"We":[25,49],"explore":[26],"from":[29,40],"Internal":[30],"Feedback":[31],"(RLIF),":[32],"a":[33,57,124],"framework":[34],"that":[35,55,85,113],"enables":[36],"LLMs":[37],"to":[38,97,127],"learn":[39],"intrinsic":[41,114],"signals":[42,116],"without":[43,103],"external":[44,68],"rewards":[45,69,135],"or":[46,107],"labeled":[47],"data.":[48],"propose":[50],"Intuitor,":[51],"an":[52],"RLIF":[53],"method":[54],"uses":[56],"model's":[58],"own":[59],"confidence-termed":[60],"self-certainty-as":[61],"its":[62],"sole":[63],"reward":[64],"signal.":[65],"Intuitor":[66,86],"replaces":[67],"in":[70],"Group":[71],"Relative":[72],"Policy":[73],"Optimization":[74],"(GRPO)":[75],"self-certainty":[77],"scores,":[78],"enabling":[79],"fully":[80],"unsupervised":[81],"learning.":[82],"Experiments":[83],"demonstrate":[84],"matches":[87],"GRPO's":[88],"performance":[89],"mathematical":[91],"benchmarks":[92],"while":[93],"achieving":[94],"better":[95],"generalization":[96],"out-of-domain":[98],"tasks":[99],"like":[100],"code":[101],"generation,":[102],"requiring":[104],"gold":[105],"solutions":[106],"test":[108],"cases.":[109],"Our":[110],"findings":[111],"show":[112],"model":[115],"can":[117],"drive":[118],"learning":[120],"across":[121],"domains,":[122],"offering":[123],"scalable":[125],"alternative":[126],"RLVR":[128],"autonomous":[130],"AI":[131],"systems":[132],"where":[133],"verifiable":[134],"are":[136],"unavailable.":[137],"Code":[138],"available":[140],"at":[141],"https://github.com/sunblaze-ucb/Intuitor":[142]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2025-10-15T00:00:00"}