{"id":"https://openalex.org/W4416767397","doi":"https://doi.org/10.21203/rs.3.rs-8196796/v1","title":"Reinforcement Learning for Large Language Model Fine-Tuning: A Systematic Literature Review","display_name":"Reinforcement Learning for Large Language Model Fine-Tuning: A Systematic Literature Review","publication_year":2025,"publication_date":"2025-11-27","ids":{"openalex":"https://openalex.org/W4416767397","doi":"https://doi.org/10.21203/rs.3.rs-8196796/v1"},"language":"en","primary_location":{"id":"doi:10.21203/rs.3.rs-8196796/v1","doi":"https://doi.org/10.21203/rs.3.rs-8196796/v1","is_oa":true,"landing_page_url":"https://doi.org/10.21203/rs.3.rs-8196796/v1","pdf_url":"https://www.researchsquare.com/article/rs-8196796/latest.pdf","source":{"id":"https://openalex.org/S4306525896","display_name":"Research Square","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":"Research Square","raw_type":"posted-content"},"volume":null,"type":"preprint","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://www.researchsquare.com/article/rs-8196796/latest.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5135310104","display_name":"Lingxiao Kong","orcid":"https://orcid.org/0009-0003-1968-7025","observed_orcids":["https://orcid.org/0009-0003-1968-7025"]},"institutions":[{"id":"https://openalex.org/I4210110507","display_name":"Fraunhofer Institute for Applied Information Technology","ror":"https://ror.org/01ak24c12","country_code":"DE","type":"facility","lineage":["https://openalex.org/I4210110507","https://openalex.org/I4923324"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Lingxiao Kong","raw_affiliation_strings":["Fraunhofer Institute for Applied Information Technology FIT"],"raw_orcid":"https://orcid.org/0009-0003-1968-7025","affiliations":[{"raw_affiliation_string":"Fraunhofer Institute for Applied Information Technology FIT","institution_ids":["https://openalex.org/I4210110507"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Qusai Ramdan","orcid":"https://orcid.org/0000-0001-8159-918X","observed_orcids":["https://orcid.org/0000-0001-8159-918X"]},"institutions":[{"id":"https://openalex.org/I177969490","display_name":"University of Southern Denmark","ror":"https://ror.org/03yrrjy16","country_code":"DK","type":"education","lineage":["https://openalex.org/I177969490"]}],"countries":["DK"],"is_corresponding":false,"raw_author_name":"Qusai Ramdan","raw_affiliation_strings":["University of Southern Denmark, Denmark"],"raw_orcid":"https://orcid.org/0000-0001-8159-918X","affiliations":[{"raw_affiliation_string":"University of Southern Denmark, Denmark","institution_ids":["https://openalex.org/I177969490"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5020876662","display_name":"Oussama Zoubia","orcid":"https://orcid.org/0000-0002-7930-7157","observed_orcids":["https://orcid.org/0000-0002-7930-7157"]},"institutions":[{"id":"https://openalex.org/I180923762","display_name":"University of Cologne","ror":"https://ror.org/00rcxh774","country_code":"DE","type":"education","lineage":["https://openalex.org/I180923762"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Oussama Zoubia","raw_affiliation_strings":["University of Cologne"],"raw_orcid":"https://orcid.org/0000-0002-7930-7157","affiliations":[{"raw_affiliation_string":"University of Cologne","institution_ids":["https://openalex.org/I180923762"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5116235515","display_name":"Jahid Hasan Polash","orcid":"https://orcid.org/0009-0000-0099-9238","observed_orcids":["https://orcid.org/0009-0000-0099-9238"]},"institutions":[{"id":"https://openalex.org/I4210110507","display_name":"Fraunhofer Institute for Applied Information Technology","ror":"https://ror.org/01ak24c12","country_code":"DE","type":"facility","lineage":["https://openalex.org/I4210110507","https://openalex.org/I4923324"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Jahid Hasan Polash","raw_affiliation_strings":["Fraunhofer Institute for Applied Information Technology FIT"],"raw_orcid":"https://orcid.org/0009-0000-0099-9238","affiliations":[{"raw_affiliation_string":"Fraunhofer Institute for Applied Information Technology FIT","institution_ids":["https://openalex.org/I4210110507"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134882360","display_name":"Mayra Roxana Elwes","orcid":"https://orcid.org/0009-0005-9454-7174","observed_orcids":["https://orcid.org/0009-0005-9454-7174"]},"institutions":[{"id":"https://openalex.org/I180923762","display_name":"University of Cologne","ror":"https://ror.org/00rcxh774","country_code":"DE","type":"education","lineage":["https://openalex.org/I180923762"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Mayra Elwes","raw_affiliation_strings":["University of Cologne"],"raw_orcid":"https://orcid.org/0009-0005-9454-7174","affiliations":[{"raw_affiliation_string":"University of Cologne","institution_ids":["https://openalex.org/I180923762"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5077658725","display_name":"Mehdi Akbari Gurabi","orcid":"https://orcid.org/0000-0002-1734-8367","observed_orcids":["https://orcid.org/0000-0002-1734-8367"]},"institutions":[{"id":"https://openalex.org/I4210110507","display_name":"Fraunhofer Institute for Applied Information Technology","ror":"https://ror.org/01ak24c12","country_code":"DE","type":"facility","lineage":["https://openalex.org/I4210110507","https://openalex.org/I4923324"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Mehdi Akbari Gurabi","raw_affiliation_strings":["Fraunhofer Institute for Applied Information Technology FIT"],"raw_orcid":"https://orcid.org/0000-0002-1734-8367","affiliations":[{"raw_affiliation_string":"Fraunhofer Institute for Applied Information Technology FIT","institution_ids":["https://openalex.org/I4210110507"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Lu Jin","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I206895457","display_name":"University of Siegen","ror":"https://ror.org/02azyry73","country_code":"DE","type":"education","lineage":["https://openalex.org/I206895457"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Lu Jin","raw_affiliation_strings":["University of Siegen"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Siegen","institution_ids":["https://openalex.org/I206895457"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5047693689","display_name":"Ekaterina Kutafina","orcid":"https://orcid.org/0000-0002-3430-5123","observed_orcids":["https://orcid.org/0000-0002-3430-5123"]},"institutions":[{"id":"https://openalex.org/I180923762","display_name":"University of Cologne","ror":"https://ror.org/00rcxh774","country_code":"DE","type":"education","lineage":["https://openalex.org/I180923762"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Ekaterina Kutafina","raw_affiliation_strings":["University of Cologne"],"raw_orcid":"https://orcid.org/0000-0002-3430-5123","affiliations":[{"raw_affiliation_string":"University of Cologne","institution_ids":["https://openalex.org/I180923762"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5072713769","display_name":"Roman Matzutt","orcid":"https://orcid.org/0000-0002-4263-5317","observed_orcids":["https://orcid.org/0000-0002-4263-5317"]},"institutions":[{"id":"https://openalex.org/I4210110507","display_name":"Fraunhofer Institute for Applied Information Technology","ror":"https://ror.org/01ak24c12","country_code":"DE","type":"facility","lineage":["https://openalex.org/I4210110507","https://openalex.org/I4923324"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Roman Matzutt","raw_affiliation_strings":["Fraunhofer Institute for Applied Information Technology FIT"],"raw_orcid":"https://orcid.org/0000-0002-4263-5317","affiliations":[{"raw_affiliation_string":"Fraunhofer Institute for Applied Information Technology FIT","institution_ids":["https://openalex.org/I4210110507"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5107209735","display_name":"Yuanbin Wang","orcid":"https://orcid.org/0000-0003-1856-5205","observed_orcids":["https://orcid.org/0000-0003-1856-5205"]},"institutions":[{"id":"https://openalex.org/I180923762","display_name":"University of Cologne","ror":"https://ror.org/00rcxh774","country_code":"DE","type":"education","lineage":["https://openalex.org/I180923762"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Yuanbin Wang","raw_affiliation_strings":["University of Cologne"],"raw_orcid":"https://orcid.org/0000-0003-1856-5205","affiliations":[{"raw_affiliation_string":"University of Cologne","institution_ids":["https://openalex.org/I180923762"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Junqi Xu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I3923682","display_name":"Soochow University","ror":"https://ror.org/05t8y2r12","country_code":"CN","type":"education","lineage":["https://openalex.org/I3923682"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Junqi Xu","raw_affiliation_strings":["Soochow University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Soochow University","institution_ids":["https://openalex.org/I3923682"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Oya Deniz Beyan","orcid":"https://orcid.org/0000-0001-7611-3501","observed_orcids":["https://orcid.org/0000-0001-7611-3501"]},"institutions":[{"id":"https://openalex.org/I4210110507","display_name":"Fraunhofer Institute for Applied Information Technology","ror":"https://ror.org/01ak24c12","country_code":"DE","type":"facility","lineage":["https://openalex.org/I4210110507","https://openalex.org/I4923324"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Oya Deniz Beyan","raw_affiliation_strings":["Fraunhofer Institute for Applied Information Technology FIT"],"raw_orcid":"https://orcid.org/0000-0001-7611-3501","affiliations":[{"raw_affiliation_string":"Fraunhofer Institute for Applied Information Technology FIT","institution_ids":["https://openalex.org/I4210110507"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Cong Yang","orcid":"https://orcid.org/0000-0002-8314-0935","observed_orcids":["https://orcid.org/0000-0002-8314-0935"]},"institutions":[{"id":"https://openalex.org/I3923682","display_name":"Soochow University","ror":"https://ror.org/05t8y2r12","country_code":"CN","type":"education","lineage":["https://openalex.org/I3923682"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Cong Yang","raw_affiliation_strings":["Soochow University"],"raw_orcid":"https://orcid.org/0000-0002-8314-0935","affiliations":[{"raw_affiliation_string":"Soochow University","institution_ids":["https://openalex.org/I3923682"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5052763223","display_name":"Zeyd Boukhers","orcid":"https://orcid.org/0000-0001-9778-9164","observed_orcids":["https://orcid.org/0000-0001-9778-9164","https://orcid.org/0009-0005-4567-9699"]},"institutions":[{"id":"https://openalex.org/I4210110507","display_name":"Fraunhofer Institute for Applied Information Technology","ror":"https://ror.org/01ak24c12","country_code":"DE","type":"facility","lineage":["https://openalex.org/I4210110507","https://openalex.org/I4923324"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Zeyd Boukhers","raw_affiliation_strings":["Fraunhofer Institute for Applied Information Technology FIT"],"raw_orcid":"https://orcid.org/0000-0001-9778-9164","affiliations":[{"raw_affiliation_string":"Fraunhofer Institute for Applied Information Technology FIT","institution_ids":["https://openalex.org/I4210110507"]}]}],"institutions":[],"countries_distinct_count":3,"institutions_distinct_count":5,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":1,"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/T10028","display_name":"Topic Modeling","score":0.5220000147819519,"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":0.5220000147819519,"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":0.23720000684261322,"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":0.10840000212192535,"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/llm-fine-tuning","display_name":"LLM fine-tuning","score":0.9190000295639038},{"id":"https://openalex.org/keywords/preference-alignment","display_name":"preference alignment","score":0.8240000009536743},{"id":"https://openalex.org/keywords/reward-modeling","display_name":"reward modeling","score":0.9929999709129333},{"id":"https://openalex.org/keywords/optimization-algorithms","display_name":"optimization algorithms","score":0.6639999747276306},{"id":"https://openalex.org/keywords/training-framework","display_name":"training framework","score":0.9039999842643738}],"concepts":[{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.5654000043869019},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.51419997215271},{"id":"https://openalex.org/C58642233","wikidata":"https://www.wikidata.org/wiki/Q8269924","display_name":"Taxonomy (biology)","level":2,"score":0.49140000343322754},{"id":"https://openalex.org/C189708586","wikidata":"https://www.wikidata.org/wiki/Q1504425","display_name":"Systematic review","level":3,"score":0.47620001435279846},{"id":"https://openalex.org/C539667460","wikidata":"https://www.wikidata.org/wiki/Q2414942","display_name":"Management science","level":1,"score":0.4214000105857849},{"id":"https://openalex.org/C199168358","wikidata":"https://www.wikidata.org/wiki/Q3367000","display_name":"Orchestration","level":3,"score":0.39590001106262207},{"id":"https://openalex.org/C136197465","wikidata":"https://www.wikidata.org/wiki/Q1729295","display_name":"Variety (cybernetics)","level":2,"score":0.3644999861717224},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.3472000062465668},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.34279999136924744},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.30889999866485596},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.304500013589859},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.304500013589859},{"id":"https://openalex.org/C188147891","wikidata":"https://www.wikidata.org/wiki/Q147638","display_name":"Cognitive science","level":1,"score":0.29980000853538513},{"id":"https://openalex.org/C56739046","wikidata":"https://www.wikidata.org/wiki/Q192060","display_name":"Knowledge management","level":1,"score":0.28690001368522644},{"id":"https://openalex.org/C2164484","wikidata":"https://www.wikidata.org/wiki/Q5170150","display_name":"Core (optical fiber)","level":2,"score":0.2757999897003174},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2549999952316284}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.21203/rs.3.rs-8196796/v1","doi":"https://doi.org/10.21203/rs.3.rs-8196796/v1","is_oa":true,"landing_page_url":"https://doi.org/10.21203/rs.3.rs-8196796/v1","pdf_url":"https://www.researchsquare.com/article/rs-8196796/latest.pdf","source":{"id":"https://openalex.org/S4306525896","display_name":"Research Square","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":"Research Square","raw_type":"posted-content"}],"best_oa_location":{"id":"doi:10.21203/rs.3.rs-8196796/v1","doi":"https://doi.org/10.21203/rs.3.rs-8196796/v1","is_oa":true,"landing_page_url":"https://doi.org/10.21203/rs.3.rs-8196796/v1","pdf_url":"https://www.researchsquare.com/article/rs-8196796/latest.pdf","source":{"id":"https://openalex.org/S4306525896","display_name":"Research Square","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"acceptedVersion","is_accepted":true,"is_published":false,"raw_source_name":"Research Square","raw_type":"posted-content"},"sustainable_development_goals":[],"study_designs":[{"display_name":"Systematic Review","id":"https://openalex.org/study-designs/systematic-review"}],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416767397.pdf","grobid_xml":"https://content.openalex.org/works/W4416767397.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7164408781","https://openalex.org/W4410947217","https://openalex.org/W4415252596","https://openalex.org/W7204186283","https://openalex.org/W4408305308","https://openalex.org/W7126088995","https://openalex.org/W4391591760","https://openalex.org/W4409220086","https://openalex.org/W7128247838","https://openalex.org/W4405468015"],"abstract_inverted_index":{"Abstract":[0],"Large":[1],"Language":[2],"Models":[3],"(LLMs)":[4],"have":[5],"been":[6,20],"developed":[7],"for":[8,180],"a":[9,70,78],"wide":[10],"range":[11],"of":[12,73,106,141,171],"language-based":[13],"tasks,":[14],"while":[15,176],"Reinforcement":[16],"Learning":[17],"(RL)":[18],"has":[19],"primarily":[21],"applied":[22],"to":[23,58,136],"decision-making":[24],"problems":[25],"such":[26],"as":[27],"robotics,":[28],"game":[29],"theory,":[30],"and":[31,61,110,117,120,130,157,160,169],"control":[32],"systems.":[33],"Nowadays,":[34],"these":[35,123],"two":[36],"paradigms":[37],"are":[38,55],"integrated":[39],"through":[40,132],"different":[41],"synergies.":[42],"In":[43],"this":[44],"literature":[45],"review,":[46],"we":[47,126],"focus":[48],"on":[49],"\\textit{RL4LLM":[50],"fine-tuning},":[51],"where":[52],"RL":[53,96],"techniques":[54],"systematically":[56],"leveraged":[57],"fine-tune":[59],"LLMs":[60,115],"align":[62],"them":[63],"with":[64],"various":[65],"preferences.":[66],"Our":[67,163],"review":[68],"provides":[69],"comprehensive":[71],"analysis":[72,164],"230":[74],"recent":[75,151,172],"publications,":[76],"presenting":[77],"methodological":[79,155],"taxonomy":[80],"that":[81],"organizes":[82],"current":[83],"research":[84,148,175],"into":[85],"three":[86,147],"primary":[87,124],"method":[88],"domains:":[89],"\\textit{Optimization":[90],"Algorithm},":[91],"concerning":[92],"innovation":[93,102],"in":[94,103],"core":[95],"update":[97],"rules;":[98],"\\textit{Training":[99],"Framework},":[100],"regarding":[101],"the":[104,107,167],"orchestration":[105],"training":[108],"process;":[109],"\\textit{Reward":[111],"Modeling},":[112],"addressing":[113],"how":[114],"learn":[116],"represent":[118],"preferences":[119],"feedback.":[121],"Within":[122],"domains,":[125],"further":[127],"analyze":[128],"methods":[129,152],"innovations":[131],"more":[133],"granular":[134],"categories":[135],"provide":[137],"an":[138],"in-depth":[139],"summary":[140],"RL4LLM":[142,173],"fine-tuning":[143,174],"research.":[144],"We":[145],"address":[146],"questions:":[149],"1)":[150],"overview,":[153],"2)":[154],"innovations,":[156],"3)":[158],"limitations":[159],"future":[161,181],"directions.":[162],"comprehensively":[165],"demonstrates":[166],"breadth":[168],"impact":[170],"highlighting":[177],"valuable":[178],"directions":[179],"investigation.":[182]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-10-31T12:09:11.096373","created_date":"2025-11-28T00:00:00"}
