{"id":"https://openalex.org/W4408105642","doi":"https://doi.org/10.1162/tacl_a_00735","title":"OPT-Tree: Speculative Decoding with Adaptive Draft Tree Structure","display_name":"OPT-Tree: Speculative Decoding with Adaptive Draft Tree Structure","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4408105642","doi":"https://doi.org/10.1162/tacl_a_00735"},"language":"en","primary_location":{"id":"doi:10.1162/tacl_a_00735","is_oa":true,"landing_page_url":"https://doi.org/10.1162/tacl_a_00735","pdf_url":"https://direct.mit.edu/tacl/article-pdf/doi/10.1162/tacl_a_00735/2506509/tacl_a_00735.pdf","source":{"id":"https://openalex.org/S2729999759","display_name":"Transactions of the Association for Computational Linguistics","issn_l":"2307-387X","issn":["2307-387X"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310320244","host_organization_name":"Association for Computational Linguistics","host_organization_lineage":["https://openalex.org/P4310320244"],"host_organization_lineage_names":["Association for Computational Linguistics"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Transactions of the Association for Computational Linguistics","raw_type":"journal-article"},"type":"article","indexed_in":["crossref","doaj"],"open_access":{"is_oa":true,"oa_status":"diamond","oa_url":"https://direct.mit.edu/tacl/article-pdf/doi/10.1162/tacl_a_00735/2506509/tacl_a_00735.pdf","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101971629","display_name":"Jikai Wang","orcid":"https://orcid.org/0009-0001-4918-7190"},"institutions":[{"id":"https://openalex.org/I3923682","display_name":"Soochow University","ror":"https://ror.org/05t8y2r12","country_code":"CN","type":"education","lineage":["https://openalex.org/I3923682"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Jikai Wang","raw_affiliation_strings":["Soochow University, China. risus254@gmail.com"],"affiliations":[{"raw_affiliation_string":"Soochow University, China. risus254@gmail.com","institution_ids":["https://openalex.org/I3923682"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5086910076","display_name":"Yi Su","orcid":"https://orcid.org/0000-0002-9569-6447"},"institutions":[{"id":"https://openalex.org/I3923682","display_name":"Soochow University","ror":"https://ror.org/05t8y2r12","country_code":"CN","type":"education","lineage":["https://openalex.org/I3923682"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yi Su","raw_affiliation_strings":["Soochow University, China. yisunlp@outlook.com"],"affiliations":[{"raw_affiliation_string":"Soochow University, China. yisunlp@outlook.com","institution_ids":["https://openalex.org/I3923682"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100657514","display_name":"Juntao Li","orcid":"https://orcid.org/0000-0002-6286-7529"},"institutions":[{"id":"https://openalex.org/I3923682","display_name":"Soochow University","ror":"https://ror.org/05t8y2r12","country_code":"CN","type":"education","lineage":["https://openalex.org/I3923682"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Juntao Li","raw_affiliation_strings":["Soochow University, China. ljt@suda.edu.cn"],"affiliations":[{"raw_affiliation_string":"Soochow University, China. ljt@suda.edu.cn","institution_ids":["https://openalex.org/I3923682"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5015450450","display_name":"Qingrong Xia","orcid":null},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Qingrong Xia","raw_affiliation_strings":["Huawei Cloud, China. xiaqingrong@huawei.com"],"affiliations":[{"raw_affiliation_string":"Huawei Cloud, China. xiaqingrong@huawei.com","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5050686535","display_name":"Zi Ye","orcid":"https://orcid.org/0000-0002-2995-6271"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zi Ye","raw_affiliation_strings":["Huawei Cloud, China. yezi3@huawei.com"],"affiliations":[{"raw_affiliation_string":"Huawei Cloud, China. yezi3@huawei.com","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5047258925","display_name":"Xinyu Duan","orcid":"https://orcid.org/0000-0002-6803-7964"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xinyu Duan","raw_affiliation_strings":["Huawei Cloud, China. duanxinyu@huawei.com"],"affiliations":[{"raw_affiliation_string":"Huawei Cloud, China. duanxinyu@huawei.com","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101848951","display_name":"Zhefeng Wang","orcid":"https://orcid.org/0009-0002-6952-3101"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhefeng Wang","raw_affiliation_strings":["Huawei Cloud, China. wangzhefeng@huawei.com"],"affiliations":[{"raw_affiliation_string":"Huawei Cloud, China. wangzhefeng@huawei.com","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5108856873","display_name":"Min Zhang","orcid":"https://orcid.org/0000-0001-8604-0959"},"institutions":[{"id":"https://openalex.org/I3923682","display_name":"Soochow University","ror":"https://ror.org/05t8y2r12","country_code":"CN","type":"education","lineage":["https://openalex.org/I3923682"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Min Zhang","raw_affiliation_strings":["Soochow University, China. zhangminmt@hotmail.com"],"affiliations":[{"raw_affiliation_string":"Soochow University, China. zhangminmt@hotmail.com","institution_ids":["https://openalex.org/I3923682"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":8,"corresponding_author_ids":["https://openalex.org/A5101971629"],"corresponding_institution_ids":["https://openalex.org/I3923682"],"apc_list":null,"apc_paid":null,"fwci":11.1416,"has_fulltext":true,"cited_by_count":5,"citation_normalized_percentile":{"value":0.97858548,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":96,"max":99},"biblio":{"volume":"13","issue":null,"first_page":"188","last_page":"199"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.991100013256073,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.991100013256073,"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/T11269","display_name":"Algorithms and Data Compression","score":0.941100001335144,"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/T10317","display_name":"Advanced Database Systems and Queries","score":0.9092000126838684,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7862807512283325},{"id":"https://openalex.org/keywords/tree","display_name":"Tree (set theory)","score":0.7584524154663086},{"id":"https://openalex.org/keywords/decoding-methods","display_name":"Decoding methods","score":0.6503328084945679},{"id":"https://openalex.org/keywords/tree-structure","display_name":"Tree structure","score":0.5972824096679688},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3228369951248169},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.3087623119354248},{"id":"https://openalex.org/keywords/binary-tree","display_name":"Binary tree","score":0.09974974393844604},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.07672718167304993}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7862807512283325},{"id":"https://openalex.org/C113174947","wikidata":"https://www.wikidata.org/wiki/Q2859736","display_name":"Tree (set theory)","level":2,"score":0.7584524154663086},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.6503328084945679},{"id":"https://openalex.org/C163797641","wikidata":"https://www.wikidata.org/wiki/Q2067937","display_name":"Tree structure","level":3,"score":0.5972824096679688},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3228369951248169},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3087623119354248},{"id":"https://openalex.org/C197855036","wikidata":"https://www.wikidata.org/wiki/Q380172","display_name":"Binary tree","level":2,"score":0.09974974393844604},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.07672718167304993},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1162/tacl_a_00735","is_oa":true,"landing_page_url":"https://doi.org/10.1162/tacl_a_00735","pdf_url":"https://direct.mit.edu/tacl/article-pdf/doi/10.1162/tacl_a_00735/2506509/tacl_a_00735.pdf","source":{"id":"https://openalex.org/S2729999759","display_name":"Transactions of the Association for Computational Linguistics","issn_l":"2307-387X","issn":["2307-387X"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310320244","host_organization_name":"Association for Computational Linguistics","host_organization_lineage":["https://openalex.org/P4310320244"],"host_organization_lineage_names":["Association for Computational Linguistics"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Transactions of the Association for Computational Linguistics","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1162/tacl_a_00735","is_oa":true,"landing_page_url":"https://doi.org/10.1162/tacl_a_00735","pdf_url":"https://direct.mit.edu/tacl/article-pdf/doi/10.1162/tacl_a_00735/2506509/tacl_a_00735.pdf","source":{"id":"https://openalex.org/S2729999759","display_name":"Transactions of the Association for Computational Linguistics","issn_l":"2307-387X","issn":["2307-387X"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310320244","host_organization_name":"Association for Computational Linguistics","host_organization_lineage":["https://openalex.org/P4310320244"],"host_organization_lineage_names":["Association for Computational Linguistics"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Transactions of the Association for Computational Linguistics","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G2027526133","display_name":null,"funder_award_id":"BK20220488","funder_id":"https://openalex.org/F4320322769","funder_display_name":"Natural Science Foundation of Jiangsu Province"},{"id":"https://openalex.org/G3114819178","display_name":null,"funder_award_id":"2023QNRC001","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G6311684539","display_name":null,"funder_award_id":"202204","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G6441875606","display_name":null,"funder_award_id":"62206194","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G6558958160","display_name":"Collaborative Research:  A gravitational inventory of the solar system using high precision minor planet astrometry","funder_award_id":"2206194","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G6600089284","display_name":"Collaborative Research:  AWARE -- International Linkages of Center for Intelligent Maintenance Systems on Web-Enabled and Tether-free Technologies","funder_award_id":"0220488","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G7320519987","display_name":null,"funder_award_id":"62276077","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G77284295","display_name":null,"funder_award_id":"2022048","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G8714109953","display_name":null,"funder_award_id":"QNRC001","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320321605","display_name":"Government of Jiangsu Province","ror":"https://ror.org/004svx814"},{"id":"https://openalex.org/F4320322769","display_name":"Natural Science Foundation of Jiangsu Province","ror":"https://ror.org/01h0zpd94"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4408105642.pdf","grobid_xml":"https://content.openalex.org/works/W4408105642.grobid-xml"},"referenced_works_count":38,"referenced_works":["https://openalex.org/W4229005866","https://openalex.org/W4285225959","https://openalex.org/W4286892945","https://openalex.org/W4288089799","https://openalex.org/W4289302788","https://openalex.org/W4308760184","https://openalex.org/W4319166707","https://openalex.org/W4380353763","https://openalex.org/W4384264648","https://openalex.org/W4384918448","https://openalex.org/W4385749968","https://openalex.org/W4389519819","https://openalex.org/W4389974792","https://openalex.org/W4391124766","https://openalex.org/W4391335474","https://openalex.org/W4391590701","https://openalex.org/W4391988257","https://openalex.org/W4401042997","https://openalex.org/W4402683901","https://openalex.org/W6754244489","https://openalex.org/W6769627184","https://openalex.org/W6803096969","https://openalex.org/W6811340617","https://openalex.org/W6846659131","https://openalex.org/W6847386241","https://openalex.org/W6849530321","https://openalex.org/W6852874933","https://openalex.org/W6853864807","https://openalex.org/W6854866820","https://openalex.org/W6855650390","https://openalex.org/W6859013747","https://openalex.org/W6859890896","https://openalex.org/W6860434696","https://openalex.org/W6860710830","https://openalex.org/W6861073873","https://openalex.org/W6861213248","https://openalex.org/W6861674332","https://openalex.org/W6862090782"],"related_works":["https://openalex.org/W2071459461","https://openalex.org/W2093885745","https://openalex.org/W2945403533","https://openalex.org/W2351266481","https://openalex.org/W2132547887","https://openalex.org/W2516987005","https://openalex.org/W2907786475","https://openalex.org/W2963447120","https://openalex.org/W4289082820","https://openalex.org/W2092152400"],"abstract_inverted_index":{"Abstract":[0],"Autoregressive":[1],"language":[2],"models":[3,30],"demonstrate":[4],"excellent":[5],"performance":[6],"in":[7,50,118,165],"various":[8],"scenarios.":[9],"However,":[10],"the":[11,29,73,105,111,115,128,146,153],"inference":[12],"efficiency":[13],"is":[14,149,156,171],"limited":[15],"by":[16],"its":[17],"one-step-one-word":[18],"generation":[19],"mode,":[20],"which":[21,64,94],"has":[22],"become":[23,31],"a":[24,37,134,166],"pressing":[25],"problem":[26],"recently":[27],"as":[28],"increasingly":[32],"larger.":[33],"Speculative":[34],"decoding":[35,120],"employs":[36],"\u201cdraft":[38],"and":[39,90,132,152],"then":[40],"verify\u201d":[41],"mechanism":[42],"to":[43,47,68,71,87,98,139],"allow":[44],"multiple":[45],"tokens":[46,164],"be":[48,96],"generated":[49],"one":[51],"step,":[52],"realizing":[53],"lossless":[54],"acceleration.":[55],"Existing":[56],"methods":[57],"mainly":[58],"adopt":[59],"fixed":[60],"heuristic":[61],"draft":[62,92,101,130,147],"structures,":[63],"do":[65],"not":[66],"adapt":[67],"different":[69],"situations":[70],"maximize":[72],"acceptance":[74,116],"length":[75,117],"during":[76],"verification.":[77],"To":[78],"alleviate":[79],"this":[80],"dilemma,":[81],"we":[82],"propose":[83],"OPT-Tree,":[84],"an":[85],"algorithm":[86],"construct":[88],"adaptive":[89],"scalable":[91],"trees,":[93],"can":[95,159],"applied":[97],"any":[99],"autoregressive":[100,143],"model.":[102],"It":[103],"searches":[104],"optimal":[106],"tree":[107],"structure":[108],"that":[109,125],"maximizes":[110],"mathematical":[112],"expectation":[113],"of":[114,137],"each":[119],"step.":[121,168],"Experimental":[122],"results":[123],"reveal":[124],"OPT-Tree":[126],"outperforms":[127],"existing":[129],"structures":[131],"achieves":[133],"speed-up":[135],"ratio":[136],"up":[138],"3.2":[140],"compared":[141],"with":[142],"decoding.":[144],"If":[145],"model":[148],"powerful":[150],"enough":[151],"node":[154],"budget":[155],"sufficient,":[157],"it":[158],"generate":[160],"more":[161],"than":[162],"ten":[163],"single":[167],"Our":[169],"code":[170],"available":[172],"at":[173],"https://github.com/Jikai0Wang/OPT-Tree.":[174]},"counts_by_year":[{"year":2026,"cited_by_count":2},{"year":2025,"cited_by_count":3}],"updated_date":"2026-04-21T08:09:41.155169","created_date":"2025-10-10T00:00:00"}