{"id":"https://openalex.org/W7136354304","doi":"https://doi.org/10.48550/arxiv.2603.12529","title":"TERMINATOR: Learning Optimal Exit Points for Early Stopping in Chain-of-Thought Reasoning","display_name":"TERMINATOR: Learning Optimal Exit Points for Early Stopping in Chain-of-Thought Reasoning","publication_year":2026,"publication_date":"2026-03-13","ids":{"openalex":"https://openalex.org/W7136354304","doi":"https://doi.org/10.48550/arxiv.2603.12529"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.12529","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.12529","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":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2603.12529","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5129598365","display_name":"Alliot Nagle","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Nagle, Alliot","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129616158","display_name":"Jakhongir Saydaliev","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Saydaliev, Jakhongir","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129490801","display_name":"Dhia Garbaya","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Garbaya, Dhia","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5063528341","display_name":"Michael Gastpar","orcid":"https://orcid.org/0000-0002-5499-5336"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gastpar, Michael","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5050142356","display_name":"Ashok Vardhan Makkuva","orcid":"https://orcid.org/0000-0001-6501-9384"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Makkuva, Ashok Vardhan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5129539595","display_name":"Hyeji Kim","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kim, Hyeji","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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.349700003862381,"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"}},"topics":[{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.349700003862381,"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/T10028","display_name":"Topic Modeling","score":0.11150000244379044,"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/T11273","display_name":"Advanced Graph Neural Networks","score":0.10450000315904617,"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/leverage","display_name":"Leverage (statistics)","score":0.765999972820282},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6933000087738037},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.5361999869346619},{"id":"https://openalex.org/keywords/point","display_name":"Point (geometry)","score":0.445499986410141},{"id":"https://openalex.org/keywords/sequence","display_name":"Sequence (biology)","score":0.3582000136375427},{"id":"https://openalex.org/keywords/underpinning","display_name":"Underpinning","score":0.3379000127315521},{"id":"https://openalex.org/keywords/commonsense-reasoning","display_name":"Commonsense reasoning","score":0.28519999980926514},{"id":"https://openalex.org/keywords/focal-point","display_name":"Focal point","score":0.28380000591278076}],"concepts":[{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.765999972820282},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6933000087738037},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6718999743461609},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.5361999869346619},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5151000022888184},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.445499986410141},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3750999867916107},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.3582000136375427},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3400000035762787},{"id":"https://openalex.org/C2780871342","wikidata":"https://www.wikidata.org/wiki/Q7883752","display_name":"Underpinning","level":2,"score":0.3379000127315521},{"id":"https://openalex.org/C193221554","wikidata":"https://www.wikidata.org/wiki/Q5153664","display_name":"Commonsense reasoning","level":2,"score":0.28519999980926514},{"id":"https://openalex.org/C2779433544","wikidata":"https://www.wikidata.org/wiki/Q1435226","display_name":"Focal point","level":3,"score":0.28380000591278076},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.2824999988079071},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.2797999978065491},{"id":"https://openalex.org/C147297375","wikidata":"https://www.wikidata.org/wiki/Q6674930","display_name":"Look-ahead","level":2,"score":0.275299996137619},{"id":"https://openalex.org/C2779458634","wikidata":"https://www.wikidata.org/wiki/Q24963715","display_name":"Debiasing","level":2,"score":0.2752000093460083},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.26910001039505005},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.26339998841285706},{"id":"https://openalex.org/C159032336","wikidata":"https://www.wikidata.org/wiki/Q2488768","display_name":"Non-monotonic logic","level":2,"score":0.26159998774528503},{"id":"https://openalex.org/C3018263672","wikidata":"https://www.wikidata.org/wiki/Q1296251","display_name":"Efficient algorithm","level":2,"score":0.2531999945640564}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.12529","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.12529","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":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2603.12529","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.12529","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":"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],"Reasoning":[1],"Models":[2],"(LRMs)":[3],"achieve":[4],"impressive":[5],"performance":[6],"on":[7,164],"complex":[8],"reasoning":[9,57,62,146],"tasks":[10],"via":[11],"Chain-of-Thought":[12],"(CoT)":[13],"reasoning,":[14],"which":[15],"enables":[16],"them":[17],"to":[18,111,139,148,191],"generate":[19],"intermediate":[20],"thinking":[21],"tokens":[22],"before":[23],"arriving":[24],"at":[25,63,109],"the":[26,42,52,121,192],"final":[27,127],"answer.":[28],"However,":[29,76],"LRMs":[30,108],"often":[31,130],"suffer":[32],"from":[33],"significant":[34,157],"overthinking,":[35],"spending":[36],"excessive":[37],"compute":[38],"time":[39],"even":[40],"after":[41],"answer":[43,128,137],"is":[44,84,119,129],"generated":[45],"early":[46],"on.":[47],"Prior":[48],"work":[49],"has":[50],"identified":[51],"existence":[53],"of":[54,124,144,162],"an":[55,104,125],"optimal":[56,78,145],"length":[58],"such":[59],"that":[60,120],"truncating":[61],"this":[64,95,100,153],"point":[65],"significantly":[66],"shortens":[67],"CoT":[68,79,160],"outputs":[69],"with":[70],"virtually":[71],"no":[72],"change":[73],"in":[74,159],"performance.":[75],"determining":[77],"lengths":[80,147,161],"for":[81,107],"practical":[82,169],"datasets":[83],"highly":[85],"non-trivial":[86],"as":[87],"they":[88],"are":[89],"fully":[90],"task":[91],"and":[92,101,132,175,182],"model-dependent.":[93],"In":[94],"paper,":[96],"we":[97,133],"precisely":[98],"address":[99],"design":[102],"Terminator,":[103],"early-exit":[105],"strategy":[106],"inference":[110,184],"mitigate":[112],"overthinking.":[113],"The":[114],"central":[115],"idea":[116],"underpinning":[117],"Terminator":[118,155],"first":[122,136],"arrival":[123],"LRM's":[126],"predictable,":[131],"leverage":[134],"these":[135],"positions":[138],"create":[140],"a":[141],"novel":[142],"dataset":[143],"train":[149],"Terminator.":[150],"Powered":[151],"by":[152,186],"approach,":[154],"achieves":[156],"reductions":[158],"14%-55%":[163],"average":[165],"across":[166],"four":[167],"challenging":[168],"datasets:":[170],"MATH-500,":[171],"AIME":[172],"2025,":[173],"HumanEval,":[174],"GPQA,":[176],"while":[177],"outperforming":[178],"current":[179],"state-of-the-art":[180],"methods":[181],"reducing":[183],"latency":[185],"more":[187],"than":[188],"2x":[189],"compared":[190],"original":[193],"LRM.":[194]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-03-17T00:00:00"}