{"id":"https://openalex.org/W7166810593","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1897","title":"Decoupled Reasoning with Implicit Fact Tokens (DRIFT): A Dual-Model Framework for Efficient Long-Context Inference","display_name":"Decoupled Reasoning with Implicit Fact Tokens (DRIFT): A Dual-Model Framework for Efficient Long-Context Inference","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166810593","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1897"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.1897","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1897","pdf_url":"https://aclanthology.org/2026.findings-acl.1897.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: ACL 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-acl.1897.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139837900","display_name":"Wenxuan Xie","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wenxuan Xie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139724042","display_name":"Xuhong Wang","orcid":null},"institutions":[{"id":"https://openalex.org/I4210100255","display_name":"Beijing Academy of Artificial Intelligence","ror":"https://ror.org/016a74861","country_code":"CN","type":"other","lineage":["https://openalex.org/I4210100255"]},{"id":"https://openalex.org/I4391012619","display_name":"Shanghai Artificial Intelligence Laboratory","ror":"https://ror.org/03wkvpx79","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4391012619"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xuhong Wang","raw_affiliation_strings":["Shanghai Artificial Intelligence Laboratory , Shanghai , China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai Artificial Intelligence Laboratory , Shanghai , China","institution_ids":["https://openalex.org/I4210100255","https://openalex.org/I4391012619"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"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":0.77495552,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"38047","last_page":"38067"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.17329999804496765,"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/T10028","display_name":"Topic Modeling","score":0.17329999804496765,"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/T11303","display_name":"Bayesian Modeling and Causal Inference","score":0.1444000005722046,"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/T11010","display_name":"Logic, Reasoning, and Knowledge","score":0.1103999987244606,"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/inference","display_name":"Inference","score":0.5062999725341797},{"id":"https://openalex.org/keywords/non-monotonic-logic","display_name":"Non-monotonic logic","score":0.30079999566078186},{"id":"https://openalex.org/keywords/stability","display_name":"Stability (learning theory)","score":0.29820001125335693},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.2669000029563904}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.640999972820282},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5062999725341797},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4893999993801117},{"id":"https://openalex.org/C159032336","wikidata":"https://www.wikidata.org/wiki/Q2488768","display_name":"Non-monotonic logic","level":2,"score":0.30079999566078186},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2985000014305115},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.29820001125335693},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.26910001039505005},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.2669000029563904},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.24959999322891235},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.24040000140666962}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.1897","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1897","pdf_url":"https://aclanthology.org/2026.findings-acl.1897.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: ACL 2026","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.1897","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1897","pdf_url":"https://aclanthology.org/2026.findings-acl.1897.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: ACL 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[{"display_name":"Reduced inequalities","score":0.4292435944080353,"id":"https://metadata.un.org/sdg/10"}],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166810593.pdf","grobid_xml":"https://content.openalex.org/works/W7166810593.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0],"integration":[1],"of":[2,20,51,146],"extensive,":[3],"dynamic":[4],"knowledge":[5,35,67,80],"into":[6,87,99],"Large":[7],"Language":[8],"Models":[9],"(LLMs)":[10],"remains":[11],"a":[12,59,78,132],"significant":[13],"challenge":[14],"due":[15],"to":[16,33,64,82],"the":[17,49,70,93,100,139],"inherent":[18],"entanglement":[19],"factual":[21],"data":[22],"and":[23,134,143],"reasoning":[24,71,101,144],"patterns.Existing":[25],"solutions,":[26],"ranging":[27],"from":[28,69],"non-parametric":[29],"Retrieval-Augmented":[30],"Generation":[31],"(RAG)":[32],"parametric":[34],"editing,":[36],"are":[37,97],"often":[38],"constrained":[39],"in":[40],"practice":[41],"by":[42],"finite":[43],"context":[44,141],"windows,":[45],"retriever":[46],"noise,":[47],"or":[48],"risk":[50],"catastrophic":[52],"forgetting.In":[53],"this":[54],"paper,":[55],"we":[56],"propose":[57],"DRIFT,":[58],"novel":[60],"dual-model":[61],"architecture":[62],"designed":[63],"explicitly":[65],"decouple":[66],"extraction":[68],"process.Unlike":[72],"static":[73],"prompt":[74],"compression,":[75],"DRIFT":[76,116],"employs":[77],"lightweight":[79],"model":[81],"dynamically":[83],"compress":[84],"document":[85],"chunks":[86],"implicit":[88],"fact":[89],"tokens":[90],"conditioned":[91],"on":[92,120],"query.These":[94],"dense":[95],"representations":[96],"projected":[98],"model's":[102],"embedding":[103],"space,":[104],"replacing":[105],"raw,":[106],"redundant":[107],"text":[108],"while":[109],"maintaining":[110],"inference":[111],"accuracy.Extensive":[112],"experiments":[113],"show":[114],"that":[115],"significantly":[117],"improves":[118],"performance":[119],"long-context":[121],"tasks,":[122],"outperforming":[123],"strong":[124],"baselines":[125],"among":[126],"comparably":[127],"sized":[128],"models.Our":[129],"approach":[130],"provides":[131],"scalable":[133],"efficient":[135],"paradigm":[136],"for":[137],"extending":[138],"effective":[140],"window":[142],"capabilities":[145],"LLMs.":[147]},"counts_by_year":[],"updated_date":"2026-08-23T07:36:19.812096","created_date":"2026-07-02T00:00:00"}