{"id":"https://openalex.org/W7140125293","doi":"https://doi.org/10.18653/v1/2026.findings-eacl.185","title":"ThinkPilot: Steering Reasoning Models via Automated Think-prefixes Optimization","display_name":"ThinkPilot: Steering Reasoning Models via Automated Think-prefixes Optimization","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7140125293","doi":"https://doi.org/10.18653/v1/2026.findings-eacl.185"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-eacl.185","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-eacl.185","pdf_url":"https://aclanthology.org/2026.findings-eacl.185.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: EACL 2026","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-eacl.185.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5130391310","display_name":"Sunzhu Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sunzhu Li","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130378146","display_name":"Zhiyu Lin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhiyu Lin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125544387","display_name":"Jiale Zhao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiale Zhao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130363939","display_name":"Shuling Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shuling Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5130371244","display_name":"Chen Wei","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen Wei","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":true,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"3573","last_page":"3592"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11596","display_name":"Constraint Satisfaction and Optimization","score":0.15800000727176666,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T11596","display_name":"Constraint Satisfaction and Optimization","score":0.15800000727176666,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T11010","display_name":"Logic, Reasoning, and Knowledge","score":0.15610000491142273,"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/T10906","display_name":"AI-based Problem Solving and Planning","score":0.09989999979734421,"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/automation","display_name":"Automation","score":0.29919999837875366},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.2759999930858612},{"id":"https://openalex.org/keywords/stability","display_name":"Stability (learning theory)","score":0.2621000111103058},{"id":"https://openalex.org/keywords/work","display_name":"Work (physics)","score":0.2540999948978424},{"id":"https://openalex.org/keywords/minification","display_name":"Minification","score":0.2529999911785126}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5799999833106995},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3578999936580658},{"id":"https://openalex.org/C115901376","wikidata":"https://www.wikidata.org/wiki/Q184199","display_name":"Automation","level":2,"score":0.29919999837875366},{"id":"https://openalex.org/C133731056","wikidata":"https://www.wikidata.org/wiki/Q4917288","display_name":"Control engineering","level":1,"score":0.28679999709129333},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.2759999930858612},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.2630999982357025},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.2621000111103058},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.2540999948978424},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.2529999911785126},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.24240000545978546}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-eacl.185","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-eacl.185","pdf_url":"https://aclanthology.org/2026.findings-eacl.185.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: EACL 2026","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-eacl.185","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-eacl.185","pdf_url":"https://aclanthology.org/2026.findings-eacl.185.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: EACL 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7140125293.pdf","grobid_xml":"https://content.openalex.org/works/W7140125293.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"Reasoning":[1],"Models":[2],"(LRMs)":[3],"are":[4,17],"powerful,":[5],"but":[6],"they":[7],"still":[8],"suffer":[9],"from":[10,88],"inefficient":[11],"and":[12,92,113,126],"off-target":[13],"reasoning.Currently,":[14],"training-free":[15,33],"methods":[16],"limited":[18],"to":[19,44,58,90],"either":[20],"rigid":[21],"heuristics":[22],"or":[23],"descriptive,":[24],"non-actionable":[25],"analyses.In":[26],"this":[27],"paper,":[28],"we":[29],"introduce":[30],"ThinkPilot,":[31],"a":[32,53,132],"framework":[34,134],"that":[35,49,105,114],"automatically":[36,124],"optimizes":[37],"LRMs":[38,137],"reasoning.It":[39],"uses":[40],"an":[41],"evolutionary":[42],"process":[43],"generate":[45],"think-prefixes,":[46],"namely":[47],"instructions":[48],"evolve":[50],"driven":[51],"by":[52],"taxonomy":[54],"of":[55,86],"reasoning":[56,111,138],"behaviors":[57],"guide":[59],"models":[60],"toward":[61],"superior":[62],"performance.Extensive":[63],"experiments":[64],"demonstrate":[65],"ThinkPilot's":[66],"broad":[67],"effectiveness:":[68],"it":[69],"significantly":[70],"improves":[71,79],"the":[72,83],"accuracy-length":[73],"tradeoff":[74],"for":[75,120,135],"efficient":[76],"reasoning,":[77],"drastically":[78],"safety":[80],"(e.g.,":[81],"cutting":[82],"StrongREJECT":[84],"score":[85],"DeepSeek-R1-Distill-Qwen-32B":[87],"27.0%":[89],"0.7%),":[91],"enhances":[93],"instruction":[94],"following.It":[95],"also":[96],"synergizes":[97],"with":[98,139],"existing":[99],"training-based":[100],"methods.Specially,":[101],"our":[102],"analysis":[103],"reveals":[104],"think-prefixes":[106],"can":[107],"reliably":[108],"control":[109],"LRMs'":[110],"behaviors,":[112,129],"different":[115],"tasks":[116],"have":[117],"strong":[118],"preferences":[119],"specific":[121],"behavioral":[122],"distributions.By":[123],"identifying":[125],"eliciting":[127],"these":[128],"ThinkPilot":[130],"provides":[131],"generalizable":[133],"aligning":[136],"task":[140],"demands.":[141]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2026-03-24T00:00:00"}