{"id":"https://openalex.org/W4408355869","doi":"https://doi.org/10.1109/icassp49660.2025.10889322","title":"Adaptive Large Language Models via Attention Shortcuts","display_name":"Adaptive Large Language Models via Attention Shortcuts","publication_year":2025,"publication_date":"2025-03-12","ids":{"openalex":"https://openalex.org/W4408355869","doi":"https://doi.org/10.1109/icassp49660.2025.10889322"},"language":"en","primary_location":{"id":"doi:10.1109/icassp49660.2025.10889322","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49660.2025.10889322","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5017296738","display_name":"Prateek Verma","orcid":"https://orcid.org/0000-0002-6402-8948"},"institutions":[{"id":"https://openalex.org/I97018004","display_name":"Stanford University","ror":"https://ror.org/00f54p054","country_code":"US","type":"education","lineage":["https://openalex.org/I97018004"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Prateek Verma","raw_affiliation_strings":["Stanford University,Department of Electrical Engineering,Stanford,CA,94305"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Stanford University,Department of Electrical Engineering,Stanford,CA,94305","institution_ids":["https://openalex.org/I97018004"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5001436196","display_name":"Mert Pilanc\u0131","orcid":"https://orcid.org/0000-0002-0870-9992"},"institutions":[{"id":"https://openalex.org/I97018004","display_name":"Stanford University","ror":"https://ror.org/00f54p054","country_code":"US","type":"education","lineage":["https://openalex.org/I97018004"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Mert Pilanci","raw_affiliation_strings":["Stanford University,Department of Electrical Engineering,Stanford,CA,94305"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Stanford University,Department of Electrical Engineering,Stanford,CA,94305","institution_ids":["https://openalex.org/I97018004"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I97018004"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"5"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9897000193595886,"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":0.9897000193595886,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.97079998254776,"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/computer-science","display_name":"Computer science","score":0.7873326539993286},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.4828410744667053},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4014471769332886},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.40041229128837585},{"id":"https://openalex.org/keywords/cognitive-science","display_name":"Cognitive science","score":0.3564915060997009},{"id":"https://openalex.org/keywords/psychology","display_name":"Psychology","score":0.15531817078590393}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7873326539993286},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4828410744667053},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4014471769332886},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.40041229128837585},{"id":"https://openalex.org/C188147891","wikidata":"https://www.wikidata.org/wiki/Q147638","display_name":"Cognitive science","level":1,"score":0.3564915060997009},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.15531817078590393}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icassp49660.2025.10889322","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49660.2025.10889322","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.44999998807907104,"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education"}],"awards":[],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320337345","display_name":"Office of Naval Research","ror":"https://ror.org/00rk2pe57"},{"id":"https://openalex.org/F4320338281","display_name":"Army Research Office","ror":"https://ror.org/05epdh915"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":25,"referenced_works":["https://openalex.org/W1494198834","https://openalex.org/W2946359678","https://openalex.org/W2946417913","https://openalex.org/W2963088785","https://openalex.org/W3174290991","https://openalex.org/W4375869259","https://openalex.org/W4381786045","https://openalex.org/W4385245566","https://openalex.org/W4385807453","https://openalex.org/W4390041933","https://openalex.org/W4397049097","https://openalex.org/W6755182157","https://openalex.org/W6760601182","https://openalex.org/W6772383348","https://openalex.org/W6773855435","https://openalex.org/W6778883912","https://openalex.org/W6784559596","https://openalex.org/W6793801364","https://openalex.org/W6810081322","https://openalex.org/W6810311916","https://openalex.org/W6843290783","https://openalex.org/W6846857581","https://openalex.org/W6848735303","https://openalex.org/W6852503157","https://openalex.org/W6855387343"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W3204019825"],"abstract_inverted_index":{"Transformer":[0],"architectures":[1],"are":[2,119],"the":[3,6,15,48,56,64,77,111,127],"backbone":[4],"of":[5,20,55],"modern":[7],"AI":[8],"revolution.":[9],"However,":[10],"these":[11],"models":[12,112],"simply":[13],"stack":[14],"same":[16],"blocks":[17],"in":[18,121],"dozens":[19],"layers":[21,58,117],"and":[22,39,80,93,96,123],"process":[23],"information":[24],"sequentially":[25],"from":[26],"one":[27],"block":[28],"to":[29,36,51,53],"another.":[30],"In":[31],"this":[32,38],"paper,":[33],"we":[34,97],"propose":[35],"challenge":[37],"introduce":[40],"adaptive":[41,120],"computations":[42],"for":[43,101],"LLM-like":[44],"setups,":[45],"which":[46],"allow":[47],"final":[49],"layer":[50],"attend":[52],"all":[54],"intermediate":[57],"as":[59],"it":[60],"deems":[61],"fit":[62],"through":[63],"attention":[65,70,108],"mechanism,":[66],"thereby":[67],"introducing":[68],"computational":[69],"shortcuts.":[71],"These":[72],"shortcuts":[73],"can":[74],"thus":[75],"make":[76],"architecture":[78],"depth":[79,124],"context":[81,122],"adaptive.":[82],"We":[83,104],"showcase":[84],"four":[85],"different":[86],"datasets,":[87],"namely":[88],"acoustic":[89],"tokens,":[90],"natural":[91],"language,":[92],"symbolic":[94],"music,":[95],"achieve":[98],"superior":[99],"performance":[100],"GPT-like":[102],"architecture.":[103],"give":[105],"evidence":[106],"via":[107],"maps":[109],"that":[110,118],"learn":[113],"complex":[114],"dependencies":[115],"across":[116],"depending":[125],"on":[126],"input":[128],"tokens.":[129]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}