{"id":"https://openalex.org/W7150772163","doi":"https://doi.org/10.48550/arxiv.2604.03098","title":"Co-Evolution of Policy and Internal Reward for Language Agents","display_name":"Co-Evolution of Policy and Internal Reward for Language Agents","publication_year":2026,"publication_date":"2026-04-03","ids":{"openalex":"https://openalex.org/W7150772163","doi":"https://doi.org/10.48550/arxiv.2604.03098"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.03098","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.03098","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2604.03098","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5133057318","display_name":"Xinyu Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Wang, Xinyu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5022531438","display_name":"Hanwei Wu","orcid":"https://orcid.org/0000-0002-3876-110X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Hanwei","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133045102","display_name":"Jingwei Song","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Song, Jingwei","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133013567","display_name":"Shuyuan Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Shuyuan","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133045538","display_name":"Jiayi Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Jiayi","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133054237","display_name":"Fanqi Kong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kong, Fanqi","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133044714","display_name":"Tung Sum Thomas Kwok","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kwok, Tung Sum Thomas","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5062087251","display_name":"Xiao-Wen Chang","orcid":"https://orcid.org/0000-0001-9251-2959"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chang, Xiao-Wen","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133062910","display_name":"Yuyu Luo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Luo, Yuyu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133016720","display_name":"Chenglin Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Chenglin","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5133001446","display_name":"Bang Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Bang","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":11,"corresponding_author_ids":["https://openalex.org/A5133057318"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.25859999656677246,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.25859999656677246,"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":0.09160000085830688,"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/T10028","display_name":"Topic Modeling","score":0.07720000296831131,"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/action","display_name":"Action (physics)","score":0.5867000222206116},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5687999725341797},{"id":"https://openalex.org/keywords/internal-model","display_name":"Internal model","score":0.544700026512146},{"id":"https://openalex.org/keywords/policy-learning","display_name":"Policy learning","score":0.4471000134944916},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.3675000071525574},{"id":"https://openalex.org/keywords/control","display_name":"Control (management)","score":0.3521000146865845}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6553999781608582},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.5867000222206116},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5687999725341797},{"id":"https://openalex.org/C28427503","wikidata":"https://www.wikidata.org/wiki/Q13580300","display_name":"Internal model","level":3,"score":0.544700026512146},{"id":"https://openalex.org/C2779436431","wikidata":"https://www.wikidata.org/wiki/Q30672407","display_name":"Policy learning","level":2,"score":0.4471000134944916},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4311000108718872},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3675000071525574},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.3521000146865845},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.34360000491142273},{"id":"https://openalex.org/C2779843651","wikidata":"https://www.wikidata.org/wiki/Q7390335","display_name":"SIGNAL (programming language)","level":2,"score":0.3167000114917755},{"id":"https://openalex.org/C180747234","wikidata":"https://www.wikidata.org/wiki/Q23373","display_name":"Cognitive psychology","level":1,"score":0.29980000853538513},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.28060001134872437},{"id":"https://openalex.org/C143661069","wikidata":"https://www.wikidata.org/wiki/Q670713","display_name":"Reward system","level":2,"score":0.26440000534057617},{"id":"https://openalex.org/C2983448237","wikidata":"https://www.wikidata.org/wiki/Q1078276","display_name":"Language understanding","level":2,"score":0.25679999589920044}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.03098","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.03098","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.03098","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.03098","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1,58,154],"model":[2],"(LLM)":[3],"agents":[4,59,155],"learn":[5],"by":[6,16,160,166],"interacting":[7],"with":[8,136,146],"environments,":[9],"but":[10,164],"long-horizon":[11],"training":[12],"remains":[13],"fundamentally":[14],"bottlenecked":[15],"sparse":[17],"and":[18,42,65,85,110,133,170,178],"delayed":[19],"rewards.":[20],"Existing":[21],"methods":[22],"typically":[23],"address":[24],"this":[25],"challenge":[26],"through":[27],"post-hoc":[28],"credit":[29],"assignment":[30],"or":[31],"external":[32],"reward":[33,45,56,93,135,175],"models,":[34],"which":[35],"provide":[36],"limited":[37],"guidance":[38,64,112],"at":[39],"inference":[40],"time":[41],"often":[43],"separate":[44],"improvement":[46],"from":[47],"policy":[48,96,106,115,132],"improvement.":[49],"We":[50],"propose":[51],"Self-Guide,":[52],"a":[53,74,102],"self-generated":[54],"internal":[55,92,117,134,174],"for":[57,94],"that":[60,153],"supports":[61],"both":[62],"inference-time":[63,123],"training-time":[66],"supervision.":[67],"Specifically,":[68],"the":[69,80,87],"agent":[70,121],"uses":[71],"Self-Guide":[72],"as":[73,116],"short":[75],"self-guidance":[76,124],"signal":[77,89],"to":[78,168],"steer":[79],"next":[81],"action":[82],"during":[83,98,176],"inference,":[84],"converts":[86],"same":[88],"into":[90],"step-level":[91],"denser":[95],"optimization":[97],"training.":[99],"This":[100],"creates":[101],"co-evolving":[103],"loop:":[104],"better":[105,108,111],"produces":[107],"guidance,":[109],"further":[113,139],"improves":[114],"reward.":[118,148],"Across":[119],"three":[120],"benchmarks,":[122],"already":[125],"yields":[126],"clear":[127],"gains,":[128],"while":[129],"jointly":[130],"evolving":[131],"GRPO":[137],"brings":[138],"improvements":[140],"(8\\%)":[141],"over":[142],"baselines":[143],"trained":[144],"solely":[145],"environment":[147],"Overall,":[149],"our":[150],"results":[151],"suggest":[152],"can":[156],"improve":[157],"not":[158],"only":[159],"collecting":[161],"more":[162],"experience,":[163],"also":[165],"learning":[167],"generate":[169],"refine":[171],"their":[172],"own":[173],"acting":[177],"learning.":[179]},"counts_by_year":[],"updated_date":"2026-04-07T06:06:30.997549","created_date":"2026-04-07T00:00:00"}