{"id":"https://openalex.org/W7148985879","doi":"https://doi.org/10.48550/arxiv.2604.01413","title":"Adaptive Stopping for Multi-Turn LLM Reasoning","display_name":"Adaptive Stopping for Multi-Turn LLM Reasoning","publication_year":2026,"publication_date":"2026-04-01","ids":{"openalex":"https://openalex.org/W7148985879","doi":"https://doi.org/10.48550/arxiv.2604.01413"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.01413","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.01413","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":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2604.01413","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5132895750","display_name":"Xiaofan Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Xiaofan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132865523","display_name":"Huy Nguyen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Nguyen, Huy","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132840714","display_name":"Bo Yu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yu, Bo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132917739","display_name":"Chenxi Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Chenxi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5132832725","display_name":"Lu Cheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cheng, Lu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"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":0.8396999835968018,"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":0.8396999835968018,"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/T13274","display_name":"Expert finding and Q&A systems","score":0.03689999878406525,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T13629","display_name":"Text Readability and Simplification","score":0.016300000250339508,"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/metric","display_name":"Metric (unit)","score":0.6000000238418579},{"id":"https://openalex.org/keywords/heuristic","display_name":"Heuristic","score":0.5838000178337097},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.5712000131607056},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.564300000667572},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.5619999766349792},{"id":"https://openalex.org/keywords/early-stopping","display_name":"Early stopping","score":0.40880000591278076},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.3479999899864197}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7635999917984009},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.6000000238418579},{"id":"https://openalex.org/C173801870","wikidata":"https://www.wikidata.org/wiki/Q201413","display_name":"Heuristic","level":2,"score":0.5838000178337097},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.5712000131607056},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.564300000667572},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.5619999766349792},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4406999945640564},{"id":"https://openalex.org/C5465570","wikidata":"https://www.wikidata.org/wiki/Q5326898","display_name":"Early stopping","level":3,"score":0.40880000591278076},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.39309999346733093},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3479999899864197},{"id":"https://openalex.org/C99414536","wikidata":"https://www.wikidata.org/wiki/Q7098950","display_name":"Optimal stopping","level":2,"score":0.32850000262260437},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3109000027179718},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.3066999912261963},{"id":"https://openalex.org/C2780898871","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Performance metric","level":2,"score":0.30230000615119934},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.30219998955726624},{"id":"https://openalex.org/C110251889","wikidata":"https://www.wikidata.org/wiki/Q1569697","display_name":"Model checking","level":2,"score":0.30090001225471497},{"id":"https://openalex.org/C80478641","wikidata":"https://www.wikidata.org/wiki/Q195771","display_name":"Sequential analysis","level":2,"score":0.2831999957561493},{"id":"https://openalex.org/C44291984","wikidata":"https://www.wikidata.org/wiki/Q1074173","display_name":"Question answering","level":2,"score":0.2721000015735626},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.27149999141693115},{"id":"https://openalex.org/C75606506","wikidata":"https://www.wikidata.org/wiki/Q1049183","display_name":"Formal methods","level":2,"score":0.2689000070095062}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.01413","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.01413","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":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.01413","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.01413","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":"Preprint"},"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":[1,130],"Models":[2,131],"(LLMs)":[3],"increasingly":[4],"rely":[5,47],"on":[6,48,165,176],"multi-turn":[7,118,141],"reasoning":[8],"and":[9,17,56,81,88,115,168,179,192,207],"interaction,":[10],"such":[11,78],"as":[12,79],"adaptive":[13,121,166],"retrieval-augmented":[14],"generation":[15],"(RAG)":[16],"ReAct-style":[18],"agents,":[19],"to":[20,110,153],"answer":[21],"difficult":[22],"questions.":[23],"These":[24],"methods":[25,107],"improve":[26],"accuracy":[27],"by":[28],"iteratively":[29],"retrieving":[30],"information,":[31],"reasoning,":[32],"or":[33,52],"acting,":[34],"but":[35,104],"introduce":[36,198],"a":[37,111,199],"key":[38],"challenge:":[39],"\\textbf{When":[40],"should":[41],"the":[42,62,67,136,151,173,186],"model":[43,113,152],"stop?}":[44],"Existing":[45],"approaches":[46],"heuristic":[49],"stopping":[50,91],"rules":[51],"fixed":[53],"turn":[54],"budgets":[55,147],"provide":[57],"no":[58],"formal":[59,101],"guarantees":[60],"that":[61,202],"final":[63],"prediction":[64,98,193],"still":[65],"contains":[66],"correct":[68],"answer.":[69],"This":[70],"limitation":[71],"is":[72],"particularly":[73],"problematic":[74],"in":[75],"high-stakes":[76],"domains":[77],"finance":[80],"healthcare,":[82],"where":[83,170],"unnecessary":[84],"turns":[85],"increase":[86],"cost":[87],"latency,":[89],"while":[90,156,184],"too":[92],"early":[93,155],"risks":[94],"incorrect":[95],"decisions.":[96],"Conformal":[97,133],"(CP)":[99],"provides":[100],"coverage":[102,160,175,205],"guarantees,":[103],"existing":[105],"LLM-CP":[106],"only":[108],"apply":[109],"single":[112],"output":[114],"cannot":[116],"handle":[117],"pipelines":[119],"with":[120,132],"stopping.":[122],"To":[123],"address":[124],"this":[125],"gap,":[126],"we":[127],"propose":[128],"Multi-Turn":[129],"Prediction":[134],"(MiCP),":[135],"first":[137],"CP":[138],"framework":[139],"for":[140],"reasoning.":[142],"MiCP":[143,164],"allocates":[144],"different":[145],"error":[146],"across":[148],"turns,":[149,189],"enabling":[150],"stop":[154],"maintaining":[157],"an":[158],"overall":[159],"guarantee.":[161],"We":[162,196],"demonstrate":[163],"RAG":[167],"ReAct,":[169],"it":[171],"achieves":[172],"target":[174],"both":[177],"single-hop":[178],"multi-hop":[180],"question":[181],"answering":[182,208],"benchmarks":[183],"reducing":[185],"number":[187],"of":[188],"inference":[190],"cost,":[191],"set":[194],"size.":[195],"further":[197],"new":[200],"metric":[201],"jointly":[203],"evaluates":[204],"validity":[206],"efficiency.":[209]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-04-04T00:00:00"}