{"id":"https://openalex.org/W4416034093","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.894","title":"TwT: Thinking without Tokens by Habitual Reasoning Distillation with Multi-Teachers\u2019 Guidance","display_name":"TwT: Thinking without Tokens by Habitual Reasoning Distillation with Multi-Teachers\u2019 Guidance","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416034093","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.894"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.894","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.894","pdf_url":"https://aclanthology.org/2025.findings-emnlp.894.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.894.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5024662712","display_name":"Jingxian Xu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jingxian Xu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101603009","display_name":"Mengyu Zhou","orcid":"https://orcid.org/0000-0002-0322-7513"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mengyu Zhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5022424132","display_name":"Weichang Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Weichang Liu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Hanbing Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hanbing Liu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5049068599","display_name":"Shi Han","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shi Han","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5100331492","display_name":"Dongmei Zhang","orcid":"https://orcid.org/0000-0003-2497-3678"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dongmei Zhang","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":1,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"16475","last_page":"16489"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11902","display_name":"Intelligent Tutoring Systems and Adaptive Learning","score":0.22220000624656677,"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/T11902","display_name":"Intelligent Tutoring Systems and Adaptive Learning","score":0.22220000624656677,"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":0.16269999742507935,"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/T11516","display_name":"Visual and Cognitive Learning Processes","score":0.07739999890327454,"subfield":{"id":"https://openalex.org/subfields/3205","display_name":"Experimental and Cognitive Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.32919999957084656},{"id":"https://openalex.org/keywords/case-based-reasoning","display_name":"Case-based reasoning","score":0.28859999775886536},{"id":"https://openalex.org/keywords/distillation","display_name":"Distillation","score":0.267300009727478},{"id":"https://openalex.org/keywords/automated-reasoning","display_name":"Automated reasoning","score":0.24959999322891235},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.24650000035762787}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5958999991416931},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.529699981212616},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.32919999957084656},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.29030001163482666},{"id":"https://openalex.org/C20162079","wikidata":"https://www.wikidata.org/wiki/Q1151406","display_name":"Case-based reasoning","level":2,"score":0.28859999775886536},{"id":"https://openalex.org/C204030448","wikidata":"https://www.wikidata.org/wiki/Q101017","display_name":"Distillation","level":2,"score":0.267300009727478},{"id":"https://openalex.org/C195344581","wikidata":"https://www.wikidata.org/wiki/Q2555318","display_name":"Automated reasoning","level":2,"score":0.24959999322891235},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.24650000035762787},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.23839999735355377},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.22509999573230743}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-emnlp.894","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.894","pdf_url":"https://aclanthology.org/2025.findings-emnlp.894.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.894","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.894","pdf_url":"https://aclanthology.org/2025.findings-emnlp.894.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/W4416034093.pdf","grobid_xml":"https://content.openalex.org/works/W4416034093.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"have":[4],"made":[5],"significant":[6],"strides":[7],"in":[8,19,129],"problem-solving":[9],"by":[10,81],"incorporating":[11],"reasoning":[12,16,50,69],"processes.However,":[13],"this":[14,34],"enhanced":[15],"capability":[17],"results":[18,111],"an":[20],"increased":[21],"number":[22],"of":[23],"output":[24,133],"tokens":[25,134],"during":[26],"inference,":[27],"leading":[28],"to":[29,125,136],"higher":[30],"computational":[31],"costs.To":[32],"address":[33],"challenge,":[35],"we":[36,84],"propose":[37,85],"TwT":[38,114],"(Thinking":[39],"without":[40],"Tokens),":[41],"a":[42,61,76,90,94,126,141],"method":[43,106],"that":[44,92,113],"reduces":[45,116],"inference-time":[46],"costs":[47,118],"through":[48,75],"habitual":[49,73],"distillation":[51,98,138],"with":[52,131],"multi-teachers'":[53],"guidance,":[54],"while":[55,119],"maintaining":[56],"high":[57],"performance.Our":[58],"approach":[59],"introduces":[60],"Habitual":[62],"Reasoning":[63],"Distillation":[64],"method,":[65],"which":[66],"internalizes":[67],"explicit":[68],"into":[70],"the":[71],"model's":[72],"behavior":[74],"Teacher-Guided":[77],"compression":[78],"strategy":[79],"inspired":[80],"human":[82],"cognition.Additionally,":[83],"Dual-Criteria":[86],"Rejection":[87],"Sampling":[88],"(DCRS),":[89],"technique":[91],"generates":[93],"highquality":[95],"and":[96],"diverse":[97],"dataset":[99],"using":[100],"multiple":[101],"teacher":[102],"models,":[103],"making":[104],"our":[105],"suitable":[107],"for":[108,145],"unsupervised":[109],"scenarios.Experimental":[110],"demonstrate":[112],"effectively":[115],"inference":[117],"preserving":[120],"superior":[121],"performance,":[122],"achieving":[123],"up":[124],"13.6%":[127],"improvement":[128],"accuracy":[130],"fewer":[132],"compared":[135],"other":[137],"methods,":[139],"offering":[140],"highly":[142],"practical":[143],"solution":[144],"efficient":[146],"LLM":[147],"deployment.":[148]},"counts_by_year":[{"year":2025,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-11-08T00:00:00"}