{"id":"https://openalex.org/W4399304331","doi":"https://doi.org/10.1145/3650200.3656619","title":"Sylva: Sparse Embedded Adapters via Hierarchical Approximate Second-Order Information","display_name":"Sylva: Sparse Embedded Adapters via Hierarchical Approximate Second-Order Information","publication_year":2024,"publication_date":"2024-05-30","ids":{"openalex":"https://openalex.org/W4399304331","doi":"https://doi.org/10.1145/3650200.3656619"},"language":"en","primary_location":{"id":"doi:10.1145/3650200.3656619","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3650200.3656619","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3650200.3656619","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 38th ACM International Conference on Supercomputing","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://dl.acm.org/doi/pdf/10.1145/3650200.3656619","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5025040062","display_name":"Baorun Mu","orcid":"https://orcid.org/0000-0002-5447-205X"},"institutions":[{"id":"https://openalex.org/I185261750","display_name":"University of Toronto","ror":"https://ror.org/03dbr7087","country_code":"CA","type":"education","lineage":["https://openalex.org/I185261750"]},{"id":"https://openalex.org/I4210127509","display_name":"Vector Institute","ror":"https://ror.org/03kqdja62","country_code":"CA","type":"facility","lineage":["https://openalex.org/I4210127509"]}],"countries":["CA"],"is_corresponding":true,"raw_author_name":"Baorun Mu","raw_affiliation_strings":["University of Toronto, Vector Institute, CentML, Canada"],"affiliations":[{"raw_affiliation_string":"University of Toronto, Vector Institute, CentML, Canada","institution_ids":["https://openalex.org/I4210127509","https://openalex.org/I185261750"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5044162748","display_name":"Christina Giannoula","orcid":"https://orcid.org/0000-0003-0162-4547"},"institutions":[{"id":"https://openalex.org/I185261750","display_name":"University of Toronto","ror":"https://ror.org/03dbr7087","country_code":"CA","type":"education","lineage":["https://openalex.org/I185261750"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Christina Giannoula","raw_affiliation_strings":["University of Toronto, CentML, Canada"],"affiliations":[{"raw_affiliation_string":"University of Toronto, CentML, Canada","institution_ids":["https://openalex.org/I185261750"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5073370044","display_name":"Shang Wang","orcid":"https://orcid.org/0000-0001-5150-0353"},"institutions":[{"id":"https://openalex.org/I185261750","display_name":"University of Toronto","ror":"https://ror.org/03dbr7087","country_code":"CA","type":"education","lineage":["https://openalex.org/I185261750"]},{"id":"https://openalex.org/I4210127509","display_name":"Vector Institute","ror":"https://ror.org/03kqdja62","country_code":"CA","type":"facility","lineage":["https://openalex.org/I4210127509"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Shang Wang","raw_affiliation_strings":["University of Toronto, Vector Institute, CentML, Canada"],"affiliations":[{"raw_affiliation_string":"University of Toronto, Vector Institute, CentML, Canada","institution_ids":["https://openalex.org/I4210127509","https://openalex.org/I185261750"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5007585346","display_name":"Gennady Pekhimenko","orcid":"https://orcid.org/0000-0002-3839-0919"},"institutions":[{"id":"https://openalex.org/I185261750","display_name":"University of Toronto","ror":"https://ror.org/03dbr7087","country_code":"CA","type":"education","lineage":["https://openalex.org/I185261750"]},{"id":"https://openalex.org/I4210127509","display_name":"Vector Institute","ror":"https://ror.org/03kqdja62","country_code":"CA","type":"facility","lineage":["https://openalex.org/I4210127509"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Gennady Pekhimenko","raw_affiliation_strings":["University of Toronto, Vector Institute, CentML, Canada"],"affiliations":[{"raw_affiliation_string":"University of Toronto, Vector Institute, CentML, Canada","institution_ids":["https://openalex.org/I4210127509","https://openalex.org/I185261750"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":4,"corresponding_author_ids":["https://openalex.org/A5025040062"],"corresponding_institution_ids":["https://openalex.org/I185261750","https://openalex.org/I4210127509"],"apc_list":null,"apc_paid":null,"fwci":0.3504,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":{"value":0.50909863,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":91,"max":95},"biblio":{"volume":null,"issue":null,"first_page":"485","last_page":"497"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11447","display_name":"Blind Source Separation Techniques","score":0.9994000196456909,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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/T11447","display_name":"Blind Source Separation Techniques","score":0.9994000196456909,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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/T10320","display_name":"Neural Networks and Applications","score":0.9976000189781189,"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/T12357","display_name":"Digital Media Forensic Detection","score":0.9909999966621399,"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/computer-science","display_name":"Computer science","score":0.5561180114746094},{"id":"https://openalex.org/keywords/order","display_name":"Order (exchange)","score":0.5479488372802734},{"id":"https://openalex.org/keywords/theoretical-computer-science","display_name":"Theoretical computer science","score":0.3536564111709595},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.34226831793785095},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.3369186520576477}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5561180114746094},{"id":"https://openalex.org/C182306322","wikidata":"https://www.wikidata.org/wiki/Q1779371","display_name":"Order (exchange)","level":2,"score":0.5479488372802734},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3536564111709595},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.34226831793785095},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.3369186520576477},{"id":"https://openalex.org/C10138342","wikidata":"https://www.wikidata.org/wiki/Q43015","display_name":"Finance","level":1,"score":0.0},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3650200.3656619","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3650200.3656619","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3650200.3656619","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 38th ACM International Conference on Supercomputing","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3650200.3656619","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3650200.3656619","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3650200.3656619","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 38th ACM International Conference on Supercomputing","raw_type":"proceedings-article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","score":0.7099999785423279,"display_name":"Quality Education"}],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":false},"content_urls":{"pdf":"https://content.openalex.org/works/W4399304331.pdf"},"referenced_works_count":16,"referenced_works":["https://openalex.org/W1970789124","https://openalex.org/W2004026774","https://openalex.org/W2154249783","https://openalex.org/W2474694648","https://openalex.org/W2560674852","https://openalex.org/W2784747162","https://openalex.org/W2954698171","https://openalex.org/W2981852735","https://openalex.org/W3081168214","https://openalex.org/W3086499488","https://openalex.org/W3129831491","https://openalex.org/W3204998121","https://openalex.org/W3205803342","https://openalex.org/W4288089799","https://openalex.org/W4311887664","https://openalex.org/W4312933868"],"related_works":["https://openalex.org/W2748952813","https://openalex.org/W4391375266","https://openalex.org/W1979597421","https://openalex.org/W2007980826","https://openalex.org/W2061531152","https://openalex.org/W3002753104","https://openalex.org/W2077600819","https://openalex.org/W2142036596","https://openalex.org/W2072657027","https://openalex.org/W2600246793"],"abstract_inverted_index":{"Fine-tuning":[0],"is":[1,248],"the":[2,33,51,60,69,76,81,90,99,129,144,150,172,182,213,221,226,235],"gateway":[3],"to":[4,75,88,187,233,240],"transferring":[5],"learned":[6],"knowledge":[7],"in":[8,55,137,225],"a":[9,39,107,138,155],"pre-trained":[10,34,56,183],"Large":[11],"Language":[12],"Model":[13],"(LLM)":[14],"on":[15,124,200],"many":[16],"downstream":[17,125],"applications.":[18,126],"To":[19,97],"make":[20],"LLM":[21,35,109,132,184],"fine-tuning":[22,110,118,186,196],"more":[23],"affordable,":[24],"prior":[25],"works":[26],"follow":[27],"two":[28],"paths:":[29],"i)":[30,154],"adapters":[31],"freeze":[32],"weights":[36,44,54,63,92,133,175,180],"and":[37,47,58,93,119,141,146,161,166,176,207],"inject":[38],"small":[40],"number":[41],"of":[42,101,149,158,246],"trainable":[43,78],"during":[45,64,117,185],"fine-tuning,":[46],"ii)":[48,162],"pruners":[49],"remove":[50],"less":[52],"important":[53,91,131,174],"LLMs":[57],"train":[59],"remaining":[61],"sparse":[62,173,179],"fine-tuning.":[65],"We":[66,192],"find":[67],"that":[68,112,194],"former":[70],"introduces":[71,83],"computation":[72,145],"overheads":[73],"due":[74],"injected":[77],"parameters,":[79],"while":[80],"latter":[82],"an":[84,163],"expensive":[85],"pre-processing":[86,139,151,227],"step":[87,152,228],"identify":[89],"degrades":[94],"model":[95,122],"quality.":[96],"get":[98],"best":[100],"both":[102],"worlds,":[103],"we":[104],"propose":[105],"Sylva,":[106],"novel":[108],"procedure":[111],"provides":[113],"high":[114,189],"system":[115,190],"performance":[116],"attains":[120],"state-of-the-art":[121,214],"quality":[123],"Sylva":[127,169,198,247],"identifies":[128],"most":[130,236],"via":[134,153],"second-order":[135,159,241],"information":[136],"step,":[140],"significantly":[142],"reduces":[143,220],"storage":[147],"costs":[148],"hierarchical":[156,218],"approximation":[157,219,239],"information,":[160],"online":[164],"projection":[165],"rediagonalization":[167],"algorithm.":[168],"trains":[170],"only":[171],"embeds":[177],"these":[178],"into":[181],"provide":[188],"performance.":[191],"show":[193],"end-to-end":[195],"with":[197],"is,":[199],"average,":[201],"5.1":[202],"\u00d7":[203,209,231],"faster":[204,210],"than":[205,211],"ZeRO":[206],"1.2":[208],"LoRA,":[212],"adapter":[215],"approach.":[216],"Sylva\u2019s":[217],"peak":[222],"GPU":[223],"memory":[224],"by":[229],"2.3":[230],"compared":[232],"K-FAC,":[234],"widely":[237],"used":[238],"information.":[242],"The":[243],"source":[244],"code":[245],"publicly":[249],"available":[250],"at":[251],"https://github.com/CentML/Sylva.":[252]},"counts_by_year":[{"year":2025,"cited_by_count":1}],"updated_date":"2026-03-27T05:58:40.876381","created_date":"2025-10-10T00:00:00"}