{"id":"https://openalex.org/W4409480953","doi":"https://doi.org/10.14428/esann/2025.es2025-97","title":"Early Prediction of Dynamic Sparsity in Large Language Models","display_name":"Early Prediction of Dynamic Sparsity in Large Language Models","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4409480953","doi":"https://doi.org/10.14428/esann/2025.es2025-97"},"language":"en","primary_location":{"id":"doi:10.14428/esann/2025.es2025-97","is_oa":true,"landing_page_url":"https://doi.org/10.14428/esann/2025.es2025-97","pdf_url":"https://doi.org/10.14428/esann/2025.es2025-97","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ESANN 2025 proceesdings","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.14428/esann/2025.es2025-97","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5025323483","display_name":"Reza Sedghi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Reza Sedghi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5090904233","display_name":"Amit Pal","orcid":"https://orcid.org/0000-0003-2979-8127"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Amit Kumar Pal","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5016175782","display_name":"Anand Subramoney","orcid":"https://orcid.org/0000-0002-7333-9860"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Anand Subramoney","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5052732444","display_name":"David Kappel","orcid":"https://orcid.org/0000-0001-5942-4033"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"David Kappel","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":null,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"651","last_page":"656"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9265999794006348,"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":0.9265999794006348,"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/T10028","display_name":"Topic Modeling","score":0.9045000076293945,"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":0.7632821798324585},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4139666259288788},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.38000935316085815}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7632821798324585},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4139666259288788},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.38000935316085815}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.14428/esann/2025.es2025-97","is_oa":true,"landing_page_url":"https://doi.org/10.14428/esann/2025.es2025-97","pdf_url":"https://doi.org/10.14428/esann/2025.es2025-97","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ESANN 2025 proceesdings","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.14428/esann/2025.es2025-97","is_oa":true,"landing_page_url":"https://doi.org/10.14428/esann/2025.es2025-97","pdf_url":"https://doi.org/10.14428/esann/2025.es2025-97","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ESANN 2025 proceesdings","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G7560871968","display_name":null,"funder_award_id":"16ME0733","funder_id":"https://openalex.org/F4320321114","funder_display_name":"Bundesministerium f\u00fcr Bildung und Forschung"}],"funders":[{"id":"https://openalex.org/F4320321114","display_name":"Bundesministerium f\u00fcr Bildung und Forschung","ror":"https://ror.org/04pz7b180"},{"id":"https://openalex.org/F4320331625","display_name":"Gauss Centre for Supercomputing","ror":"https://ror.org/0585fsj26"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4409480953.pdf","grobid_xml":"https://content.openalex.org/works/W4409480953.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"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":{"Large":[0],"language":[1],"models":[2],"are":[3],"powerful":[4],"but":[5],"computationally":[6],"very":[7],"expensive.We":[8],"investigate":[9],"dynamic":[10,25,71,91],"sparsity":[11,44,72],"in":[12,29,78,85],"attention":[13,30,51],"mechanisms,":[14],"using":[15],"the":[16,24,37,48,79,88],"OPT":[17],"model":[18,38,80],"as":[19,53],"a":[20,54,96],"case":[21],"study.We":[22],"explore":[23],"nature":[26],"of":[27,36,50],"redundancy":[28],"heads":[31],"and":[32,66,101],"analyze":[33],"which":[34],"components":[35],"provide":[39],"sufficient":[40],"information":[41],"to":[42],"predict":[43],"effectively.Our":[45],"findings":[46],"highlight":[47],"norm":[49],"outputs":[52],"reliable":[55],"criterion":[56],"for":[57,98],"ranking":[58],"head":[59],"importance.We":[60],"systematically":[61],"evaluate":[62],"embeddings":[63],"across":[64],"layers":[65],"time":[67],"steps,":[68],"showing":[69],"that":[70],"predictions":[73],"can":[74],"be":[75],"achieved":[76],"early":[77],"pipeline":[81],"with":[82],"minimal":[83],"loss":[84],"accuracy.By":[86],"elucidating":[87],"mechanisms":[89],"underlying":[90],"sparsity,":[92],"this":[93],"work":[94],"lays":[95],"foundation":[97],"more":[99],"efficient":[100],"scalable":[102],"transformer":[103],"models.":[104]},"counts_by_year":[{"year":2025,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}