{"id":"https://openalex.org/W4416034283","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.928","title":"Faster and Better LLMs via Latency-Aware Test-Time Scaling","display_name":"Faster and Better LLMs via Latency-Aware Test-Time Scaling","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416034283","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.928"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.928","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.928","pdf_url":"https://aclanthology.org/2025.findings-emnlp.928.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-emnlp.928.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100693880","display_name":"Zili Wang","orcid":"https://orcid.org/0000-0002-5003-3092"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zili Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100425837","display_name":"Tianyu Zhang","orcid":"https://orcid.org/0000-0001-8632-3510"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tianyu Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5080943673","display_name":"Haoli Bai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Haoli Bai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5079270380","display_name":"Lu Hou","orcid":"https://orcid.org/0000-0003-3085-9353"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lu Hou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5055511467","display_name":"Xiaoqiang Yu","orcid":"https://orcid.org/0009-0004-9703-5249"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xianzhi Yu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5052415035","display_name":"W.M. Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wulong Liu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5076932826","display_name":"Shiming Xiang","orcid":"https://orcid.org/0000-0001-8184-6566"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shiming Xiang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5041397265","display_name":"Lei Zhu","orcid":"https://orcid.org/0000-0003-3826-3220"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lei Zhu","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":1.2703,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":{"value":0.81697487,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":91,"max":95},"biblio":{"volume":null,"issue":null,"first_page":"17124","last_page":"17137"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10743","display_name":"Software Testing and Debugging Techniques","score":0.2757999897003174,"subfield":{"id":"https://openalex.org/subfields/1712","display_name":"Software"},"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/T10743","display_name":"Software Testing and Debugging Techniques","score":0.2757999897003174,"subfield":{"id":"https://openalex.org/subfields/1712","display_name":"Software"},"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/T11032","display_name":"VLSI and Analog Circuit Testing","score":0.08070000261068344,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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":0.0697999969124794,"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/scaling","display_name":"Scaling","score":0.46720001101493835},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.3111000061035156},{"id":"https://openalex.org/keywords/scale","display_name":"Scale (ratio)","score":0.28360000252723694},{"id":"https://openalex.org/keywords/constraint","display_name":"Constraint (computer-aided design)","score":0.24549999833106995}],"concepts":[{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.46720001101493835},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4163999855518341},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.3111000061035156},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.31060001254081726},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.28360000252723694},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.2831999957561493},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.24660000205039978},{"id":"https://openalex.org/C2776036281","wikidata":"https://www.wikidata.org/wiki/Q48769818","display_name":"Constraint (computer-aided design)","level":2,"score":0.24549999833106995},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.2328999936580658},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.22930000722408295}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-emnlp.928","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.928","pdf_url":"https://aclanthology.org/2025.findings-emnlp.928.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.928","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.928","pdf_url":"https://aclanthology.org/2025.findings-emnlp.928.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416034283.pdf","grobid_xml":"https://content.openalex.org/works/W4416034283.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Accuracy":[0,20],"(%)":[1,21],"9":[2],"less":[3,23],"tokens":[4],"Sequential":[5,25,29],"v.s.Parallel":[6,26],"w.r.t":[7,27],"#Output":[8],"Token":[9],"2":[10,12],"7":[11],"8":[13],"Second":[14],"76":[15],"78":[16],"80":[17],"82":[18],"84":[19],"1.6":[22],"time":[24],"Latency":[28],"scaling":[30,32,49,55,62,73,105,122],"Parallel":[31],"Figure":[33],"1:":[34],"\"Compute-optimal\"":[35],"does":[36],"not":[37],"necessarily":[38],"translate":[39],"to":[40,78,103],"\"latency-optimal\"":[41],"for":[42,65],"test-time":[43,48],"scaling.Left:":[44],"Previous":[45],"works":[46],"measure":[47],"by":[50,123],"#token":[51],"budget,":[52,71],"indicating":[53],"sequential":[54,84,104],"shows":[56],"superior":[57],"token":[58],"efficiency":[59],"than":[60,83],"parallel":[61,72,121],"(majority":[63],"voting":[64,125],"instance).Right:":[66],"When":[67],"considering":[68],"latency":[69],"as":[70],"can":[74],"be":[75],"1.6x":[76],"faster":[77],"achieve":[79],"the":[80],"same":[81],"accuracy":[82],"scaling.Experiments":[85],"are":[86],"performed":[87],"using":[88],"s1.1-32B":[89],"(Muennighoff":[90],"et":[91,97],"al.,":[92,98],"2025)":[93],"on":[94],"MATH-500":[95],"(Hendrycks":[96],"2021).The":[99],"red":[100],"curve":[101,119],"corresponds":[102],"with":[106,126],"sequence":[107,129],"length":[108],"varying":[109],"in":[110],"{1024,":[111],"2048,":[112],"4096,":[113],"8192},":[114],"via":[115],"budget":[116],"forcing.The":[117],"blue":[118],"represents":[120],"majority":[124],"a":[127],"fixed":[128],"length.":[130]},"counts_by_year":[{"year":2025,"cited_by_count":1}],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-11-08T00:00:00"}