{"id":"https://openalex.org/W7160866470","doi":"https://doi.org/10.48550/arxiv.2605.07465","title":"SEIF: Self-Evolving Reinforcement Learning for Instruction Following","display_name":"SEIF: Self-Evolving Reinforcement Learning for Instruction Following","publication_year":2026,"publication_date":"2026-05-08","ids":{"openalex":"https://openalex.org/W7160866470","doi":"https://doi.org/10.48550/arxiv.2605.07465"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.07465","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.07465","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":null,"license_id":null,"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.2605.07465","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5122913726","display_name":"Qingyu Ren","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Ren, Qingyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135881228","display_name":"Qianyu He","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"He, Qianyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5054041765","display_name":"Jiajie Zhu","orcid":"https://orcid.org/0000-0002-1930-7884"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhu, Jiajie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5039782834","display_name":"Xingzhou Chen","orcid":"https://orcid.org/0000-0003-1631-5202"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Xingzhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113112970","display_name":"Jingwen Chang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chang, Jingwen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5111019890","display_name":"Zeye Sun","orcid":"https://orcid.org/0009-0008-8449-9318"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sun, Zeye","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135831548","display_name":"Han Xia","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xia, Han","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135832300","display_name":"Fei Yu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yu, Fei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135890833","display_name":"Jiaqing Liang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liang, Jiaqing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5135844961","display_name":"Yanghua Xiao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiao, Yanghua","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":10,"corresponding_author_ids":["https://openalex.org/A5122913726"],"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/T10028","display_name":"Topic Modeling","score":0.18070000410079956,"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":0.18070000410079956,"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":0.14419999718666077,"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/T11902","display_name":"Intelligent Tutoring Systems and Adaptive Learning","score":0.08020000159740448,"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/reinforcement-learning","display_name":"Reinforcement learning","score":0.7652999758720398},{"id":"https://openalex.org/keywords/overfitting","display_name":"Overfitting","score":0.7418000102043152},{"id":"https://openalex.org/keywords/filter","display_name":"Filter (signal processing)","score":0.5213000178337097},{"id":"https://openalex.org/keywords/reinforcement","display_name":"Reinforcement","score":0.45669999718666077},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.4242999851703644}],"concepts":[{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.7652999758720398},{"id":"https://openalex.org/C22019652","wikidata":"https://www.wikidata.org/wiki/Q331309","display_name":"Overfitting","level":3,"score":0.7418000102043152},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6847000122070312},{"id":"https://openalex.org/C106131492","wikidata":"https://www.wikidata.org/wiki/Q3072260","display_name":"Filter (signal processing)","level":2,"score":0.5213000178337097},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5024999976158142},{"id":"https://openalex.org/C67203356","wikidata":"https://www.wikidata.org/wiki/Q1321905","display_name":"Reinforcement","level":2,"score":0.45669999718666077},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.4242999851703644},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.38769999146461487},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3327000141143799},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.32339999079704285},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3197000026702881},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.2547999918460846}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.07465","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.07465","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.07465","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.07465","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.6580730080604553}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Instruction":[0,59],"following":[1],"is":[2],"a":[3,61,73,106,118,127,188],"fundamental":[4],"capability":[5,15,89],"of":[6,69,96,170],"large":[7],"language":[8],"models":[9],"(LLMs),":[10],"yet":[11],"continuously":[12],"improving":[13],"this":[14],"remains":[16],"challenging.":[17],"Existing":[18],"methods":[19],"typically":[20],"rely":[21],"either":[22],"on":[23,34,180],"costly":[24],"external":[25],"supervision":[26],"from":[27],"humans":[28],"or":[29,33,111],"strong":[30,163],"teacher":[31],"models,":[32],"self-play":[35],"training":[36,176,185,195],"with":[37],"static-difficulty":[38],"instructions":[39,113],"that":[40,77,101,108,120,129,156],"cannot":[41],"evolve":[42],"as":[43],"the":[44,66,79,146,168],"model's":[45,80],"capabilities":[46],"improve.":[47],"To":[48],"address":[49],"these":[50],"limitations,":[51],"we":[52],"propose":[53],"SEIF":[54,71,94,157],"(Self-Evolving":[55],"Reinforcement":[56],"Learning":[57],"for":[58,64,133,178],"Following),":[60],"self-evolving":[62],"framework":[63],"enhancing":[65],"instruction-following":[67,81,160],"ability":[68],"LLMs.":[70],"forms":[72],"closed":[74],"self-evolution":[75,179],"loop":[76],"improves":[78,159],"ability,":[82],"where":[83],"instruction":[84],"difficulty":[85],"evolution":[86,90],"and":[87,126,138,143,153,172,199,206],"model":[88,151],"reinforce":[91],"each":[92],"other.":[93],"consists":[95],"four":[97],"roles:":[98],"an":[99,174],"Instructor":[100,137],"generates":[102],"increasingly":[103],"challenging":[104],"instructions,":[105,125],"Filter":[107],"removes":[109],"conflicting":[110],"invalid":[112],"to":[114,122,186,196],"ensure":[115],"data":[116,207],"quality,":[117],"Follower":[119,139],"learns":[121],"follow":[123],"evolved":[124],"Judger":[128],"provides":[130],"reward":[131],"signals":[132],"reinforcement":[134],"learning.":[135],"The":[136,204],"are":[140,208],"alternately":[141],"trained":[142],"co-evolve":[144],"throughout":[145],"process.":[147],"Experiments":[148],"across":[149],"multiple":[150],"scales":[152],"architectures":[154],"show":[155],"consistently":[158],"performance,":[161],"suggesting":[162],"generality.":[164],"Further":[165],"analyses":[166],"reveal":[167],"sources":[169],"improvement":[171],"identify":[173],"effective":[175],"strategy":[177],"open-ended":[181],"tasks:":[182],"sufficient":[183],"early-stage":[184],"build":[187],"solid":[189],"foundation,":[190],"followed":[191],"by":[192],"moderate":[193],"late-stage":[194],"mitigate":[197],"overfitting":[198],"achieve":[200],"better":[201],"final":[202],"performance.":[203],"code":[205],"publicly":[209],"available":[210],"at":[211],"https://github.com/Rainier-rq1/SEIF.":[212]},"counts_by_year":[],"updated_date":"2026-05-12T06:14:25.881160","created_date":"2026-05-12T00:00:00"}