{"id":"https://openalex.org/W7202346014","doi":"https://doi.org/10.48550/arxiv.2608.11967","title":"LoongReflect: Boosting Long-Horizon Reflection in Search Agents via Global Perspective Distillation","display_name":"LoongReflect: Boosting Long-Horizon Reflection in Search Agents via Global Perspective Distillation","publication_year":2026,"publication_date":"2026-08-12","ids":{"openalex":"https://openalex.org/W7202346014","doi":"https://doi.org/10.48550/arxiv.2608.11967"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.11967","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.11967","pdf_url":"https://arxiv.org/pdf/2608.11967","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","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/2608.11967","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5148894440","display_name":"Zhixin Zhang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Zhixin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148862840","display_name":"Xinke Jiang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiang, Xinke","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148851477","display_name":"Zhibang Yang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yang, Zhibang","raw_affiliation_strings":["National Engineering Research Center for Software Engineering , Peking University , Beijing , China","School of Computer Science , Peking University , Beijing , China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"National Engineering Research Center for Software Engineering , Peking University , Beijing , China","institution_ids":["https://openalex.org/I20231570"]},{"raw_affiliation_string":"School of Computer Science , Peking University , Beijing , China","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5147064727","display_name":"Weixuan Xu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Weixuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5147124842","display_name":"Guohong Qiu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qiu, Guohong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5124889641","display_name":"Xu Chu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chu, Xu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5121647741","display_name":"Junfeng Zhao","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhao, Junfeng","raw_affiliation_strings":["Key Laboratory of High Confidence Software Technologies , Ministry of Education , Beijing , China","School of Computer Science , Peking University , Beijing , China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Key Laboratory of High Confidence Software Technologies , Ministry of Education , Beijing , China","institution_ids":[]},{"raw_affiliation_string":"School of Computer Science , Peking University , Beijing , China","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5139829275","display_name":"Yasha Wang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wang, Yasha","raw_affiliation_strings":["National Engineering Research Center for Software Engineering , Peking University , Beijing , China","Peking University Information Technology Institute (Tianjin Binhai) , Tianjin , China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"National Engineering Research Center for Software Engineering , Peking University , Beijing , China","institution_ids":["https://openalex.org/I20231570"]},{"raw_affiliation_string":"Peking University Information Technology Institute (Tianjin Binhai) , Tianjin , China","institution_ids":["https://openalex.org/I20231570"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"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":0.4749000072479248,"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":0.4749000072479248,"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":0.22450000047683716,"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/T10028","display_name":"Topic Modeling","score":0.08550000190734863,"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":0.8690000176429749},{"id":"https://openalex.org/keywords/backtracking","display_name":"backtracking","score":0.9369999766349792},{"id":"https://openalex.org/keywords/teacher-student-knowledge-distillation","display_name":"teacher-student distillation","score":0.6909999847412109},{"id":"https://openalex.org/keywords/group-relative-policy-optimization","display_name":"Group Relative Policy Optimization","score":0.6629999876022339},{"id":"https://openalex.org/keywords/mathematical-reasoning","display_name":"mathematical reasoning","score":0.9599999785423279}],"concepts":[{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.7019000053405762},{"id":"https://openalex.org/C156884757","wikidata":"https://www.wikidata.org/wiki/Q798554","display_name":"Backtracking","level":2,"score":0.697700023651123},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6869000196456909},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.6371999979019165},{"id":"https://openalex.org/C65682993","wikidata":"https://www.wikidata.org/wiki/Q1056451","display_name":"Reflection (computer programming)","level":2,"score":0.5946000218391418},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.5697000026702881},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.5227000117301941},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4814999997615814},{"id":"https://openalex.org/C12713177","wikidata":"https://www.wikidata.org/wiki/Q1900281","display_name":"Perspective (graphical)","level":2,"score":0.4672999978065491},{"id":"https://openalex.org/C207024777","wikidata":"https://www.wikidata.org/wiki/Q621673","display_name":"Search tree","level":3,"score":0.4327999949455261},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.4180999994277954},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.40230000019073486},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.38960000872612},{"id":"https://openalex.org/C127705205","wikidata":"https://www.wikidata.org/wiki/Q5748245","display_name":"Heuristics","level":2,"score":0.3801000118255615},{"id":"https://openalex.org/C127162648","wikidata":"https://www.wikidata.org/wiki/Q16858953","display_name":"Channel (broadcasting)","level":2,"score":0.3540000021457672},{"id":"https://openalex.org/C113174947","wikidata":"https://www.wikidata.org/wiki/Q2859736","display_name":"Tree (set theory)","level":2,"score":0.33059999346733093},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.2840000092983246},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.26820001006126404},{"id":"https://openalex.org/C106189395","wikidata":"https://www.wikidata.org/wiki/Q176789","display_name":"Markov decision process","level":3,"score":0.2612999975681305},{"id":"https://openalex.org/C84525736","wikidata":"https://www.wikidata.org/wiki/Q831366","display_name":"Decision tree","level":2,"score":0.25200000405311584}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.11967","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.11967","pdf_url":"https://arxiv.org/pdf/2608.11967","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2608.11967","doi":"https://doi.org/10.48550/arxiv.2608.11967","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.11967","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2608.11967","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.11967","pdf_url":"https://arxiv.org/pdf/2608.11967","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","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/W7202346014.pdf","grobid_xml":"https://content.openalex.org/works/W7202346014.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7223351337","https://openalex.org/W7221817212","https://openalex.org/W7213504276","https://openalex.org/W7214727491","https://openalex.org/W7117762039","https://openalex.org/W7128545158","https://openalex.org/W6891945665","https://openalex.org/W7223010977","https://openalex.org/W4414360997","https://openalex.org/W7080127675"],"abstract_inverted_index":{"Large":[0],"language":[1],"model":[2],"agents":[3],"increasingly":[4],"rely":[5],"on":[6,206],"long-horizon":[7],"reasoning":[8,212],"to":[9,40,73,183],"solve":[10,96],"complex":[11],"tasks":[12],"involving":[13],"planning,":[14],"tool":[15],"use,":[16],"and":[17,33,37,89,122,131,147,185,210,221],"memory.":[18],"A":[19,168,188],"critical":[20],"capability":[21],"in":[22],"such":[23],"settings":[24],"is":[25,52,56],"reflection:":[26],"assessing":[27],"trajectory":[28,76,117],"progress,":[29],"identifying":[30],"missing":[31,129],"evidence":[32],"unreliable":[34,141],"intermediate":[35],"states,":[36],"deciding":[38],"whether":[39],"continue,":[41],"revise,":[42],"or":[43],"abandon":[44],"the":[45,60,74,144],"current":[46,61],"branch.":[47],"Learning":[48],"effective":[49],"reflection,":[50],"however,":[51],"challenging":[53],"because":[54],"reflection":[55,106,184],"performed":[57],"locally":[58],"within":[59],"branch,":[62],"whereas":[63],"its":[64,71],"utility":[65],"can":[66],"only":[67,86],"be":[68],"determined":[69],"by":[70],"contribution":[72],"final":[75,202],"outcome.":[77],"This":[78],"local-global":[79],"mismatch":[80],"makes":[81],"outcome-based":[82,195],"reinforcement":[83,219],"learning":[84,220],"provide":[85],"local,":[87],"sparse":[88],"delayed":[90],"supervision":[91,181],"for":[92],"reflective":[93,174],"decisions.":[94],"To":[95,153],"these,":[97],"we":[98],"propose":[99],"LoongReflect,":[100],"a":[101,108,115,149,163,177],"training":[102],"framework":[103],"that":[104],"formulates":[105],"as":[107],"memory-control":[109],"policy.":[110],"The":[111],"agent":[112],"operates":[113],"over":[114,217],"reversible":[116],"tree":[118],"using":[119,194],"explicit":[120],"reflect":[121],"backtrack":[123],"actions.":[124],"Reflection":[125],"consolidates":[126],"verified":[127],"facts,":[128],"evidence,":[130],"branch-specific":[132],"risks":[133],"into":[134],"working":[135],"memory,":[136],"while":[137],"backtracking":[138,186],"removes":[139],"an":[140],"branch":[142],"from":[143,176],"active":[145],"context":[146],"preserves":[148],"concise":[150],"corrective":[151],"lesson.":[152],"learn":[154],"this":[155],"policy,":[156],"LoongReflect":[157],"combines":[158],"two":[159],"complementary":[160],"signals":[161],"through":[162],"look-ahead,":[164],"extragradient-style":[165],"coordination":[166],"mechanism.":[167],"fast":[169],"channel":[170,190],"distills":[171],"globally":[172],"informed":[173],"behavior":[175],"privileged":[178],"teacher,":[179],"with":[180,201],"restricted":[182],"tokens.":[187],"slow":[189],"optimizes":[191],"complete":[192],"trajectories":[193],"GRPO,":[196],"aligning":[197],"local":[198],"control":[199],"decisions":[200],"task":[203],"success.":[204],"Experiments":[205],"multi-hop":[207],"retrieval-augmented":[208],"generation":[209],"mathematical":[211],"benchmarks":[213],"demonstrate":[214],"consistent":[215],"improvements":[216],"outcome-only":[218],"self-distillation":[222],"baselines.":[223]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-08-14T00:00:00"}