{"id":"https://openalex.org/W7213504276","doi":"https://doi.org/10.48550/arxiv.2609.18304","title":"Rollback the World, Keep the Reflection: Rollback-Induced Reflection for Long-Horizon LLM Agents","display_name":"Rollback the World, Keep the Reflection: Rollback-Induced Reflection for Long-Horizon LLM Agents","publication_year":2026,"publication_date":"2026-09-16","ids":{"openalex":"https://openalex.org/W7213504276","doi":"https://doi.org/10.48550/arxiv.2609.18304"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.18304","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.18304","pdf_url":"https://arxiv.org/pdf/2609.18304","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","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2609.18304","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5152009064","display_name":"Yi Yu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I37461747","display_name":"Wuhan University","ror":"https://ror.org/033vjfk17","country_code":"CN","type":"education","lineage":["https://openalex.org/I37461747"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yu, Yi","raw_affiliation_strings":["Wuhan University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Wuhan University","institution_ids":["https://openalex.org/I37461747"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5114129125","display_name":"Liuyi Yao","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yao, Liuyi","raw_affiliation_strings":["Alibaba Group"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Alibaba Group","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5152010211","display_name":"Yaliang Li","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I45928872","display_name":"Alibaba Group (China)","ror":"https://ror.org/00k642b80","country_code":"CN","type":"company","lineage":["https://openalex.org/I45928872"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Li, Yaliang","raw_affiliation_strings":["Alibaba Group"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Alibaba Group","institution_ids":["https://openalex.org/I45928872"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5085805626","display_name":"Enshu Wang","orcid":"https://orcid.org/0000-0001-7253-4763","observed_orcids":["https://orcid.org/0000-0001-7253-4763"]},"institutions":[{"id":"https://openalex.org/I37461747","display_name":"Wuhan University","ror":"https://ror.org/033vjfk17","country_code":"CN","type":"education","lineage":["https://openalex.org/I37461747"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wang, Enshu","raw_affiliation_strings":["Wuhan University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Wuhan University","institution_ids":["https://openalex.org/I37461747"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5152031210","display_name":"Libing Wu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I37461747","display_name":"Wuhan University","ror":"https://ror.org/033vjfk17","country_code":"CN","type":"education","lineage":["https://openalex.org/I37461747"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wu, Libing","raw_affiliation_strings":["Wuhan University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Wuhan University","institution_ids":["https://openalex.org/I37461747"]}]}],"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/T10906","display_name":"AI-based Problem Solving and Planning","score":0.5268999934196472,"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/T10906","display_name":"AI-based Problem Solving and Planning","score":0.5268999934196472,"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":0.36250001192092896,"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":0.019500000402331352,"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-based-agents","display_name":"LLM agents","score":0.8370000123977661},{"id":"https://openalex.org/keywords/long-horizon-tasks","display_name":"long-horizon tasks","score":0.9200000166893005},{"id":"https://openalex.org/keywords/error-recovery","display_name":"error recovery","score":0.7480000257492065},{"id":"https://openalex.org/keywords/state-restoration","display_name":"state restoration","score":0.9110000133514404},{"id":"https://openalex.org/keywords/knowledge-retention","display_name":"knowledge retention","score":0.8029999732971191}],"concepts":[{"id":"https://openalex.org/C174220543","wikidata":"https://www.wikidata.org/wiki/Q395307","display_name":"Rollback","level":3,"score":0.9556999802589417},{"id":"https://openalex.org/C65682993","wikidata":"https://www.wikidata.org/wiki/Q1056451","display_name":"Reflection (computer programming)","level":2,"score":0.7702999711036682},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7329999804496765},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.7013000249862671},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.654699981212616},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.5722000002861023},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.4700999855995178},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.4693000018596649},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.45570001006126404},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.3384999930858612},{"id":"https://openalex.org/C17020691","wikidata":"https://www.wikidata.org/wiki/Q139677","display_name":"Operator (biology)","level":5,"score":0.33660000562667847},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3086000084877014},{"id":"https://openalex.org/C2779662365","wikidata":"https://www.wikidata.org/wiki/Q5416694","display_name":"Event (particle physics)","level":2,"score":0.2888999879360199},{"id":"https://openalex.org/C2776544517","wikidata":"https://www.wikidata.org/wiki/Q189447","display_name":"Unexpected events","level":2,"score":0.2847999930381775},{"id":"https://openalex.org/C4554734","wikidata":"https://www.wikidata.org/wiki/Q593744","display_name":"Knowledge base","level":2,"score":0.27459999918937683},{"id":"https://openalex.org/C200601418","wikidata":"https://www.wikidata.org/wiki/Q2193887","display_name":"Reliability engineering","level":1,"score":0.26579999923706055},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.2502000033855438}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.18304","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.18304","pdf_url":"https://arxiv.org/pdf/2609.18304","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.2609.18304","doi":"https://doi.org/10.48550/arxiv.2609.18304","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.18304","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:2609.18304","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.18304","pdf_url":"https://arxiv.org/pdf/2609.18304","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":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7213504276.pdf","grobid_xml":"https://content.openalex.org/works/W7213504276.grobid-xml"},"referenced_works_count":10,"referenced_works":["https://openalex.org/W3126325318","https://openalex.org/W4304195432","https://openalex.org/W4389422760","https://openalex.org/W4405903187","https://openalex.org/W4414939355","https://openalex.org/W4416895257","https://openalex.org/W7077506418","https://openalex.org/W7162343841","https://openalex.org/W7164534302","https://openalex.org/W7203638561"],"related_works":["https://openalex.org/W7203638561","https://openalex.org/W7223351337","https://openalex.org/W7221520278","https://openalex.org/W7202346014","https://openalex.org/W4416036324","https://openalex.org/W7171942477","https://openalex.org/W7168577596","https://openalex.org/W7216160010","https://openalex.org/W7164160483","https://openalex.org/W7167619710"],"abstract_inverted_index":{"Large":[0],"language":[1],"model":[2],"(LLM)":[3],"agents":[4],"increasingly":[5],"tackle":[6],"long-horizon":[7,153],"tasks":[8],"through":[9,130],"multi-step":[10],"environment":[11,39],"interaction,":[12],"yet":[13],"a":[14,72,100,108,131,141],"single":[15],"erroneous":[16],"action":[17],"can":[18],"alter":[19],"subsequent":[20,124],"states":[21,40,44],"and":[22,57,85,137,147,174],"observations,":[23],"causing":[24],"errors":[25],"to":[26,52,80,83,107,122],"compound":[27],"over":[28,134],"time.":[29],"Existing":[30],"methods":[31],"either":[32],"correct":[33],"the":[34,119],"context":[35],"without":[36],"repairing":[37],"altered":[38],"or":[41],"restore":[42],"earlier":[43],"while":[45,112],"discarding":[46],"useful":[47,172],"experience,":[48],"making":[49],"it":[50],"difficult":[51],"both":[53],"eliminate":[54],"failure":[55],"conditions":[56],"avoid":[58],"repeating":[59],"past":[60],"mistakes.":[61],"We":[62,126],"argue":[63],"that":[64,76,104,156],"reliable":[65],"recovery":[66,102,129],"should":[67,88],"instead":[68],"be":[69],"treated":[70],"as":[71],"rollback-boundary":[73],"control":[74],"problem":[75],"jointly":[77],"determines":[78],"when":[79],"intervene,":[81],"where":[82],"resume,":[84],"what":[86],"information":[87],"survive":[89],"recovery.":[90,179],"Based":[91],"on":[92,151],"this":[93],"view,":[94],"we":[95],"propose":[96],"Rollback-Induced":[97],"Reflection":[98],"(RIR),":[99],"unified":[101,132],"framework":[103],"restores":[105],"execution":[106],"selected":[109],"prior":[110],"state":[111,145],"carrying":[113],"forward":[114],"reusable":[115],"knowledge":[116,148],"distilled":[117],"from":[118],"abandoned":[120],"trajectory":[121],"guide":[123],"decisions.":[125],"further":[127],"characterize":[128],"operator":[133],"rollback":[135,176],"depth":[136],"retained":[138],"memory,":[139],"providing":[140],"general":[142],"view":[143],"of":[144],"restoration":[146],"retention.":[149],"Experiments":[150],"three":[152],"benchmarks":[154],"show":[155],"RIR":[157],"consistently":[158],"improves":[159],"average":[160],"task":[161],"performance":[162],"across":[163],"multiple":[164],"LLM":[165],"backbones,":[166],"with":[167],"structured":[168],"reflection":[169],"memory":[170],"preserving":[171],"experience":[173],"selective":[175],"enabling":[177],"efficient":[178]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-09-18T00:00:00"}