{"id":"https://openalex.org/W4416034744","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.770","title":"ReCUT: Balancing Reasoning Length and Accuracy in LLMs via Stepwise Trails and Preference Optimization","display_name":"ReCUT: Balancing Reasoning Length and Accuracy in LLMs via Stepwise Trails and Preference Optimization","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416034744","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.770"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.770","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.770","pdf_url":"https://aclanthology.org/2025.findings-emnlp.770.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.770.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5073191363","display_name":"Zheyan Jin","orcid":"https://orcid.org/0000-0001-8466-7520"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhensheng Jin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5019596003","display_name":"Xinze Li","orcid":"https://orcid.org/0009-0005-7795-6574"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xinze Li","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102497859","display_name":"Yifan Ji","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yifan Ji","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5059605898","display_name":"Chunyi Peng","orcid":"https://orcid.org/0000-0003-2945-4981"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chunyi Peng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5070196864","display_name":"Zhenghao Liu","orcid":"https://orcid.org/0000-0001-8067-1477"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhenghao Liu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5115602860","display_name":"Qi Shi","orcid":"https://orcid.org/0000-0003-4727-1337"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qi Shi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5034266748","display_name":"Yukun Yan","orcid":"https://orcid.org/0000-0003-2421-7098"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yukun Yan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100400212","display_name":"Shuo Wang","orcid":"https://orcid.org/0009-0006-5388-2774"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shuo Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5006565370","display_name":"Furong Peng","orcid":"https://orcid.org/0000-0003-4461-7355"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Furong Peng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5113657466","display_name":"Yu Ge","orcid":"https://orcid.org/0000-0001-7153-5035"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ge Yu","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":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.39610727,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"14269","last_page":"14282"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11596","display_name":"Constraint Satisfaction and Optimization","score":0.2896000146865845,"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":0.2896000146865845,"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":0.09120000153779984,"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":0.05209999904036522,"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/preference","display_name":"Preference","score":0.5548999905586243},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.28189998865127563},{"id":"https://openalex.org/keywords/optimization-problem","display_name":"Optimization problem","score":0.25459998846054077},{"id":"https://openalex.org/keywords/multi-objective-optimization","display_name":"Multi-objective optimization","score":0.24050000309944153}],"concepts":[{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.5548999905586243},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5081999897956848},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4519999921321869},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2915000021457672},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.28189998865127563},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.27570000290870667},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.2732999920845032},{"id":"https://openalex.org/C137836250","wikidata":"https://www.wikidata.org/wiki/Q984063","display_name":"Optimization problem","level":2,"score":0.25459998846054077},{"id":"https://openalex.org/C42475967","wikidata":"https://www.wikidata.org/wiki/Q194292","display_name":"Operations research","level":1,"score":0.2476000040769577},{"id":"https://openalex.org/C68781425","wikidata":"https://www.wikidata.org/wiki/Q2052203","display_name":"Multi-objective optimization","level":2,"score":0.24050000309944153}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-emnlp.770","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.770","pdf_url":"https://aclanthology.org/2025.findings-emnlp.770.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.770","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.770","pdf_url":"https://aclanthology.org/2025.findings-emnlp.770.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":[{"id":"https://openalex.org/G2602313477","display_name":null,"funder_award_id":"62206042","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G3675952725","display_name":null,"funder_award_id":"N25ZLL045","funder_id":"https://openalex.org/F4320335787","funder_display_name":"Fundamental Research Funds for the Central Universities"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320335787","display_name":"Fundamental Research Funds for the Central Universities","ror":null}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416034744.pdf","grobid_xml":"https://content.openalex.org/works/W4416034744.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Recent":[0],"advances":[1],"in":[2],"Chain-of-Thought":[3],"(CoT)":[4],"prompting":[5],"have":[6],"substantially":[7],"improved":[8],"the":[9,52,55,63,79,126,138],"reasoning":[10,29,40,84,104,124,148,158,167],"capabilities":[11],"of":[12,54,83,140],"Large":[13],"Language":[14],"Models":[15],"(LLMs).However,":[16],"these":[17,141],"methods":[18],"often":[19,49],"suffer":[20],"from":[21],"overthinking,":[22],"leading":[23],"to":[24,33,60,100,111,115,170],"unnecessarily":[25],"lengthy":[26],"or":[27,165],"redundant":[28],"traces.Existing":[30],"approaches":[31],"attempt":[32],"mitigate":[34],"this":[35],"issue":[36],"through":[37],"curating":[38],"multiple":[39,146],"chains":[41],"for":[42,123,128],"training":[43],"LLMs,":[44],"but":[45],"their":[46],"effectiveness":[47],"is":[48,134],"constrained":[50],"by":[51,136,160],"quality":[53],"generated":[56],"data":[57],"and":[58,81,92,109,150],"prone":[59],"overfitting.To":[61],"address":[62],"challenge,":[64],"we":[65],"propose":[66],"Reasoning":[67],"Compression":[68],"ThroUgh":[69],"Stepwise":[70],"Trials":[71],"(ReCUT),":[72],"a":[73,88,93],"novel":[74],"method":[75],"aimed":[76],"at":[77],"balancing":[78],"accuracy":[80,168],"length":[82],"trajectory.Specifically,":[85],"Re-CUT":[86],"employs":[87],"stepwise":[89],"exploration":[90],"mechanism":[91],"long-short":[94],"switched":[95],"sampling":[96],"strategy,":[97],"enabling":[98],"LLMs":[99],"incrementally":[101],"generate":[102],"diverse":[103],"paths.These":[105],"paths":[106],"are":[107],"evaluated":[108],"used":[110],"construct":[112],"preference":[113],"pairs":[114],"train":[116],"two":[117,142],"specialized":[118],"models":[119,152],"(Gemini":[120],"LLMs)one":[121],"optimized":[122],"accuracy,":[125],"other":[127],"shorter":[129],"reasoning.A":[130],"final":[131],"integrated":[132],"model":[133],"obtained":[135],"interpolating":[137],"parameters":[139],"models.Experimental":[143],"results":[144],"across":[145],"math":[147],"datasets":[149],"backbone":[151],"demonstrate":[153],"that":[154],"ReCUT":[155],"significantly":[156],"reduces":[157],"lengths":[159],"approximately":[161],"30-50%,":[162],"while":[163],"maintaining":[164],"improving":[166],"compared":[169],"various":[171],"baselines.":[172]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-11-08T00:00:00"}