{"id":"https://openalex.org/W4412945494","doi":"https://doi.org/10.18653/v1/2025.acl-long.485","title":"Attention Entropy is a Key Factor: An Analysis of Parallel Context Encoding with Full-attention-based Pre-trained Language Models","display_name":"Attention Entropy is a Key Factor: An Analysis of Parallel Context Encoding with Full-attention-based Pre-trained Language Models","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412945494","doi":"https://doi.org/10.18653/v1/2025.acl-long.485"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.acl-long.485","doi":"https://doi.org/10.18653/v1/2025.acl-long.485","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.485","pdf_url":"https://aclanthology.org/2025.acl-long.485.pdf","source":{"id":"https://openalex.org/S4393915740","display_name":"Proceedings of the conference - Association for Computational Linguistics. Meeting","issn_l":"0736-587X","issn":["0736-587X"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":["norway-1"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"bronze","oa_url":"https://aclanthology.org/2025.acl-long.485.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101677227","display_name":"Zhisong Zhang","orcid":"https://orcid.org/0000-0002-3680-8701","observed_orcids":["https://orcid.org/0000-0002-3680-8701"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhisong Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100761418","display_name":"Yan Wang","orcid":"https://orcid.org/0000-0001-5341-6040","observed_orcids":["https://orcid.org/0000-0001-5341-6040"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yan Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102524460","display_name":"Xinting Huang","orcid":"https://orcid.org/0000-0001-6827-7426","observed_orcids":["https://orcid.org/0000-0001-6827-7426"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xinting Huang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5046409172","display_name":"Tianqing Fang","orcid":"https://orcid.org/0000-0002-0186-8253","observed_orcids":["https://orcid.org/0000-0002-0186-8253"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tianqing Fang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100445028","display_name":"Hongming Zhang","orcid":"https://orcid.org/0000-0001-8172-6484","observed_orcids":["https://orcid.org/0000-0001-8172-6484"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hongming Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Chenlong Deng","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chenlong Deng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5009237347","display_name":"Shuaiyi Li","orcid":"https://orcid.org/0009-0005-0659-676X","observed_orcids":["https://orcid.org/0009-0005-0659-676X","https://orcid.org/0009-0008-0896-9189"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shuaiyi Li","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5001067641","display_name":"Dong Yu","orcid":"https://orcid.org/0000-0002-6769-1622","observed_orcids":["https://orcid.org/0000-0002-6769-1622"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dong 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":2.8489,"has_fulltext":true,"cited_by_count":3,"citation_normalized_percentile":{"value":0.91015953,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":97,"max":98},"biblio":{"volume":null,"issue":null,"first_page":"9840","last_page":"9855"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9247999787330627,"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":0.9247999787330627,"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":0.04859999939799309,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.00419999985024333,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/attention-entropy","display_name":"attention entropy","score":0.972000002861023},{"id":"https://openalex.org/keywords/language-model-pretraining","display_name":"language model pretraining","score":0.5040000081062317},{"id":"https://openalex.org/keywords/attention-mechanism","display_name":"attention mechanism","score":0.39800000190734863}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7830970287322998},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.6609145998954773},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.6070095300674438},{"id":"https://openalex.org/C2781039887","wikidata":"https://www.wikidata.org/wiki/Q1391724","display_name":"Factor (programming language)","level":2,"score":0.5515590906143188},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.49749425053596497},{"id":"https://openalex.org/C106301342","wikidata":"https://www.wikidata.org/wiki/Q4117933","display_name":"Entropy (arrow of time)","level":2,"score":0.4785528779029846},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4395171105861664},{"id":"https://openalex.org/C183322885","wikidata":"https://www.wikidata.org/wiki/Q17007702","display_name":"Context model","level":3,"score":0.4202401041984558},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.3807704448699951},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.35119348764419556},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3477601110935211},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.15144342184066772},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.10224881768226624},{"id":"https://openalex.org/C151730666","wikidata":"https://www.wikidata.org/wiki/Q7205","display_name":"Paleontology","level":1,"score":0.0},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0},{"id":"https://openalex.org/C2781238097","wikidata":"https://www.wikidata.org/wiki/Q175026","display_name":"Object (grammar)","level":2,"score":0.0},{"id":"https://openalex.org/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","level":1,"score":0.0}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2025.acl-long.485","doi":"https://doi.org/10.18653/v1/2025.acl-long.485","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.485","pdf_url":"https://aclanthology.org/2025.acl-long.485.pdf","source":{"id":"https://openalex.org/S4393915740","display_name":"Proceedings of the conference - Association for Computational Linguistics. Meeting","issn_l":"0736-587X","issn":["0736-587X"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":["norway-1"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2412.16545","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2412.16545","pdf_url":"https://arxiv.org/pdf/2412.16545","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2412.16545","doi":"https://doi.org/10.48550/arxiv.2412.16545","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2412.16545","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.acl-long.485","doi":"https://doi.org/10.18653/v1/2025.acl-long.485","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.485","pdf_url":"https://aclanthology.org/2025.acl-long.485.pdf","source":{"id":"https://openalex.org/S4393915740","display_name":"Proceedings of the conference - Association for Computational Linguistics. Meeting","issn_l":"0736-587X","issn":["0736-587X"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":["norway-1"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412945494.pdf","grobid_xml":"https://content.openalex.org/works/W4412945494.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4417289958","https://openalex.org/W7131884702","https://openalex.org/W7154502384","https://openalex.org/W4416054714","https://openalex.org/W4389984381","https://openalex.org/W4416047809","https://openalex.org/W4413835240","https://openalex.org/W4412602684","https://openalex.org/W4403579453","https://openalex.org/W4394906145"],"abstract_inverted_index":{"Large":[0],"language":[1,12],"models":[2],"have":[3],"shown":[4],"remarkable":[5],"performance":[6,86,148],"across":[7],"a":[8,100,115],"wide":[9],"range":[10],"of":[11,26,103],"tasks,":[13],"owing":[14],"to":[15,85,123,156],"their":[16],"exceptional":[17],"capabilities":[18],"in":[19,34],"context":[20,27,61,66,158],"modeling.The":[21],"most":[22],"commonly":[23],"used":[24],"method":[25,40],"modeling":[28,159],"is":[29,59],"full":[30],"self-attention,":[31],"as":[32],"seen":[33],"standard":[35],"decoder-only":[36],"Transformers.Although":[37],"powerful,":[38],"this":[39,96,104,151],"can":[41,113,153],"be":[42,114],"inefficient":[43],"for":[44],"long":[45],"sequences":[46],"and":[47,69,91,106,131,146],"may":[48],"overlook":[49],"inherent":[50],"input":[51],"structures.To":[52],"address":[53],"these":[54,139],"problems,":[55],"an":[56],"alternative":[57],"approach":[58],"parallel":[60,73,82],"encoding,":[62],"which":[63],"splits":[64],"the":[65,88],"into":[67],"sub-pieces":[68],"encodes":[70],"them":[71],"parallelly.Because":[72],"patterns":[74],"are":[75,94],"not":[76],"encountered":[77],"during":[78],"training,":[79],"naively":[80],"applying":[81],"encoding":[83],"leads":[84],"degradation.However,":[87],"underlying":[89],"reasons":[90],"potential":[92],"mitigations":[93],"unclear.In":[95],"work,":[97],"we":[98,118],"provide":[99],"detailed":[101],"analysis":[102],"issue":[105],"identify":[107],"that":[108,138],"unusually":[109],"high":[110],"attention":[111,125,129,144],"entropy":[112,126,145],"key":[116],"factor.Furthermore,":[117],"adopt":[119],"two":[120],"straightforward":[121],"methods":[122,140],"reduce":[124],"by":[127],"incorporating":[128],"sinks":[130],"selective":[132],"mechanisms.Experiments":[133],"on":[134],"various":[135],"tasks":[136],"reveal":[137],"effectively":[141],"lower":[142],"irregular":[143],"narrow":[147],"gaps.We":[149],"hope":[150],"study":[152],"illuminate":[154],"ways":[155],"enhance":[157],"mechanisms.":[160]},"counts_by_year":[{"year":2026,"cited_by_count":3}],"updated_date":"2026-10-29T09:56:32.521200","created_date":"2025-10-10T00:00:00"}