{"id":"https://openalex.org/W7165887942","doi":"https://doi.org/10.48550/arxiv.2606.25354","title":"Efficient and Trainable Language Model Test-Time Scaling via Local Branch Routing","display_name":"Efficient and Trainable Language Model Test-Time Scaling via Local Branch Routing","publication_year":2026,"publication_date":"2026-06-24","ids":{"openalex":"https://openalex.org/W7165887942","doi":"https://doi.org/10.48550/arxiv.2606.25354"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.25354","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.25354","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.2606.25354","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139382696","display_name":"Yutong Yin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yin, Yutong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139358408","display_name":"Mingyu Jin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jin, Mingyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139340808","display_name":"Jin Pan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Pan, Jin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134964327","display_name":"Changyi Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Changyi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134976831","display_name":"Zijie Xia","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xia, Zijie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120251973","display_name":"Dhruv Pai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Pai, Dhruv","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5105753106","display_name":"Hu S","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hu, Shuming","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139343023","display_name":"Zhen Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Zhen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139355147","display_name":"Chenyang Zhao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Chenyang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139308089","display_name":"Jinman Zhao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Jinman","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139313180","display_name":"Wujiang Xu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Wujiang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5052597908","display_name":"R K Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Raymond","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139325865","display_name":"Xin Eric Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Xin Eric","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139331428","display_name":"Julian McAuley","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"McAuley, Julian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139385843","display_name":"Zhaoran Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Zhaoran","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":0.539900004863739,"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":0.539900004863739,"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":0.10270000249147415,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.09610000252723694,"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/router","display_name":"Router","score":0.6862000226974487},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.6092000007629395},{"id":"https://openalex.org/keywords/routing","display_name":"Routing (electronic design automation)","score":0.5730000138282776},{"id":"https://openalex.org/keywords/scaling","display_name":"Scaling","score":0.42410001158714294},{"id":"https://openalex.org/keywords/branching","display_name":"Branching (polymer chemistry)","score":0.39969998598098755},{"id":"https://openalex.org/keywords/verifiable-secret-sharing","display_name":"Verifiable secret sharing","score":0.3732999861240387},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.36480000615119934},{"id":"https://openalex.org/keywords/graph","display_name":"Graph","score":0.36340001225471497}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7003999948501587},{"id":"https://openalex.org/C2775896111","wikidata":"https://www.wikidata.org/wiki/Q642560","display_name":"Router","level":2,"score":0.6862000226974487},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.6092000007629395},{"id":"https://openalex.org/C74172769","wikidata":"https://www.wikidata.org/wiki/Q1446839","display_name":"Routing (electronic design automation)","level":2,"score":0.5730000138282776},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.42410001158714294},{"id":"https://openalex.org/C206175624","wikidata":"https://www.wikidata.org/wiki/Q595731","display_name":"Branching (polymer chemistry)","level":2,"score":0.39969998598098755},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.388700008392334},{"id":"https://openalex.org/C85847156","wikidata":"https://www.wikidata.org/wiki/Q59015987","display_name":"Verifiable secret sharing","level":3,"score":0.3732999861240387},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.36480000615119934},{"id":"https://openalex.org/C132525143","wikidata":"https://www.wikidata.org/wiki/Q141488","display_name":"Graph","level":2,"score":0.36340001225471497},{"id":"https://openalex.org/C184896649","wikidata":"https://www.wikidata.org/wiki/Q290066","display_name":"Routing table","level":4,"score":0.3492000102996826},{"id":"https://openalex.org/C62611344","wikidata":"https://www.wikidata.org/wiki/Q1062658","display_name":"Node (physics)","level":2,"score":0.3416000008583069},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.3366999924182892},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.3140000104904175},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.30959999561309814},{"id":"https://openalex.org/C137471899","wikidata":"https://www.wikidata.org/wiki/Q11381","display_name":"Local area network","level":2,"score":0.29350000619888306},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2881999909877777},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.2705000042915344},{"id":"https://openalex.org/C204948658","wikidata":"https://www.wikidata.org/wiki/Q1119410","display_name":"Static routing","level":4,"score":0.2705000042915344},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.26980000734329224},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.2678000032901764},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2583000063896179},{"id":"https://openalex.org/C110121322","wikidata":"https://www.wikidata.org/wiki/Q865811","display_name":"Distribution (mathematics)","level":2,"score":0.2529999911785126},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.2526000142097473}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.25354","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.25354","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.2606.25354","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.25354","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":[{"id":"https://metadata.un.org/sdg/16","display_name":"Peace, Justice and strong institutions","score":0.4392799735069275}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Test-time":[0],"scaling":[1,41],"improves":[2,170],"language-model":[3,201],"reasoning,":[4],"but":[5],"existing":[6],"approaches":[7],"often":[8],"face":[9],"a":[10,38,45,60,109],"difficult":[11],"trade-off:":[12],"long":[13],"chain-of-thought":[14],"sampling":[15],"remains":[16],"single-threaded,":[17],"while":[18,93],"sentence-":[19],"or":[20],"solution-level":[21,96],"search":[22],"can":[23],"be":[24],"computationally":[25],"expensive":[26],"and":[27,58,107,121,141,173,181,197],"hard":[28],"to":[29,63,68,85],"train":[30],"end-to-end.":[31],"We":[32],"introduce":[33],"Local":[34],"Branch":[35],"Routing":[36],"(LBR),":[37],"token-level":[39],"test-time":[40,202],"framework":[42],"that":[43,157,189],"expands":[44],"small":[46],"local":[47,78,191],"lookahead":[48],"tree,":[49],"forwards":[50],"all":[51],"sampled":[52],"branches":[53],"through":[54],"the":[55,65,73,89,138,144],"language":[56],"model,":[57],"uses":[59],"lightweight":[61,190],"router":[62,122,142],"select":[64],"depth-1":[66],"subtree":[67],"commit.":[69],"By":[70],"routing":[71,163],"over":[72,175],"hidden":[74,159],"states":[75,160],"of":[76,200],"candidate":[77],"futures,":[79],"LBR":[80,155,169],"allows":[81],"each":[82],"token":[83],"decision":[84],"use":[86],"evidence":[87],"beyond":[88],"root":[90],"next-token":[91],"distribution":[92],"avoiding":[94],"full":[95],"search.":[97],"The":[98],"resulting":[99],"prune-shift-grow":[100],"decoding":[101],"process":[102],"preserves":[103],"discrete":[104,176,198],"branch":[105],"identities":[106],"defines":[108],"tractable":[110],"tree-trajectory":[111],"likelihood:":[112],"newly":[113],"grown":[114],"nodes":[115],"are":[116,124],"counted":[117],"when":[118],"first":[119],"sampled,":[120],"decisions":[123],"assigned":[125],"explicit":[126],"probabilities.":[127],"This":[128],"enables":[129],"end-to-end":[130],"reinforcement":[131],"learning":[132],"with":[133],"verifiable":[134],"rewards,":[135],"jointly":[136],"optimizing":[137],"base":[139],"model":[140],"under":[143],"same":[145],"likelihood-ratio":[146],"principle":[147],"as":[148],"discrete-token":[149,179],"RLVR.":[150],"On":[151,165],"synthetic":[152],"hierarchical-planning":[153],"tasks,":[154],"shows":[156],"post-candidate":[158],"provide":[161],"useful":[162],"evidence.":[164],"mathematical":[166],"reasoning":[167],"benchmarks,":[168],"both":[171],"Pass@1":[172],"Pass@32":[174],"chain-of-thought,":[177],"vanilla":[178],"RLVR,":[180],"RL-compatible":[182],"soft-token":[183],"branching":[184,192],"baselines.":[185],"These":[186],"results":[187],"suggest":[188],"offers":[193],"an":[194],"efficient,":[195],"trainable,":[196],"form":[199],"scaling.":[203]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-26T00:00:00"}