{"id":"https://openalex.org/W7160198217","doi":"https://doi.org/10.1007/s44443-026-00724-4","title":"Layer-wise heterogeneity-guided heterogeneous pruning: An LLM compression method","display_name":"Layer-wise heterogeneity-guided heterogeneous pruning: An LLM compression method","publication_year":2026,"publication_date":"2026-04-30","ids":{"openalex":"https://openalex.org/W7160198217","doi":"https://doi.org/10.1007/s44443-026-00724-4"},"language":"en","primary_location":{"id":"doi:10.1007/s44443-026-00724-4","is_oa":true,"landing_page_url":"https://doi.org/10.1007/s44443-026-00724-4","pdf_url":null,"source":{"id":"https://openalex.org/S2764955546","display_name":"Journal of King Saud University - Computer and Information Sciences","issn_l":"1319-1578","issn":["1319-1578","2213-1248"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310320990","host_organization_name":"Elsevier BV","host_organization_lineage":["https://openalex.org/P4310320990"],"host_organization_lineage_names":["Elsevier BV"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Journal of King Saud University Computer and Information Sciences","raw_type":"journal-article"},"type":"article","indexed_in":["crossref","doaj"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1007/s44443-026-00724-4","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5135114788","display_name":"Zhihao Yu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhihao Yu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135265646","display_name":"Zuxin Ma","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zuxin Ma","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5086384170","display_name":"Guowei Shen","orcid":"https://orcid.org/0000-0002-1329-4058"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Guowei Shen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135263066","display_name":"Yi Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yi Chen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5126054736","display_name":"Chun Guo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chun Guo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135218416","display_name":"Xiaoqi Duan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiaoqi Duan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135163532","display_name":"Qing Qian","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qing Qian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5135123870","display_name":"Yunhe Cui","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yunhe Cui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":{"value":1990,"currency":"USD","value_usd":1990},"apc_paid":{"value":1990,"currency":"USD","value_usd":1990},"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.43369129,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"38","issue":"6","first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11269","display_name":"Algorithms and Data Compression","score":0.6158999800682068,"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/T11269","display_name":"Algorithms and Data Compression","score":0.6158999800682068,"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/T12326","display_name":"Network Packet Processing and Optimization","score":0.02539999969303608,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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":0.023600000888109207,"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/compression","display_name":"Compression (physics)","score":0.4359000027179718},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.28459998965263367},{"id":"https://openalex.org/keywords/deformation","display_name":"Deformation (meteorology)","score":0.25060001015663147},{"id":"https://openalex.org/keywords/work","display_name":"Work (physics)","score":0.24529999494552612},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.22759999334812164}],"concepts":[{"id":"https://openalex.org/C180016635","wikidata":"https://www.wikidata.org/wiki/Q2712821","display_name":"Compression (physics)","level":2,"score":0.4359000027179718},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.36800000071525574},{"id":"https://openalex.org/C192562407","wikidata":"https://www.wikidata.org/wiki/Q228736","display_name":"Materials science","level":0,"score":0.33309999108314514},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.28459998965263367},{"id":"https://openalex.org/C199639397","wikidata":"https://www.wikidata.org/wiki/Q1788588","display_name":"Engineering drawing","level":1,"score":0.25220000743865967},{"id":"https://openalex.org/C204366326","wikidata":"https://www.wikidata.org/wiki/Q3027650","display_name":"Deformation (meteorology)","level":2,"score":0.25060001015663147},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.24529999494552612},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.23340000212192535},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.22759999334812164},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.226500004529953}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1007/s44443-026-00724-4","is_oa":true,"landing_page_url":"https://doi.org/10.1007/s44443-026-00724-4","pdf_url":null,"source":{"id":"https://openalex.org/S2764955546","display_name":"Journal of King Saud University - Computer and Information Sciences","issn_l":"1319-1578","issn":["1319-1578","2213-1248"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310320990","host_organization_name":"Elsevier BV","host_organization_lineage":["https://openalex.org/P4310320990"],"host_organization_lineage_names":["Elsevier BV"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Journal of King Saud University Computer and Information Sciences","raw_type":"journal-article"},{"id":"pmh:oai:doaj.org/article:d11e754e0f4744789cf66db85d5b7415","is_oa":false,"landing_page_url":"https://doaj.org/article/d11e754e0f4744789cf66db85d5b7415","pdf_url":null,"source":{"id":"https://openalex.org/S4306401280","display_name":"DOAJ (DOAJ: Directory of Open Access Journals)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Journal of King Saud University: Computer and Information Sciences, Vol 38, Iss 6 (2026)","raw_type":"article"}],"best_oa_location":{"id":"doi:10.1007/s44443-026-00724-4","is_oa":true,"landing_page_url":"https://doi.org/10.1007/s44443-026-00724-4","pdf_url":null,"source":{"id":"https://openalex.org/S2764955546","display_name":"Journal of King Saud University - Computer and Information Sciences","issn_l":"1319-1578","issn":["1319-1578","2213-1248"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310320990","host_organization_name":"Elsevier BV","host_organization_lineage":["https://openalex.org/P4310320990"],"host_organization_lineage_names":["Elsevier BV"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Journal of King Saud University Computer and Information Sciences","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":18,"referenced_works":["https://openalex.org/W2890894339","https://openalex.org/W2910280493","https://openalex.org/W2923014074","https://openalex.org/W2946609015","https://openalex.org/W2946659172","https://openalex.org/W3194676777","https://openalex.org/W4308090748","https://openalex.org/W4313055133","https://openalex.org/W4390872847","https://openalex.org/W4399051247","https://openalex.org/W4409364145","https://openalex.org/W4409364387","https://openalex.org/W4412887764","https://openalex.org/W4412888794","https://openalex.org/W4415799245","https://openalex.org/W4415800408","https://openalex.org/W4415924966","https://openalex.org/W7133245307"],"related_works":[],"abstract_inverted_index":{"Abstract":[0],"Large":[1],"language":[2,11],"models":[3],"(LLMs)":[4],"have":[5],"become":[6,43],"foundational":[7],"in":[8,37],"modern":[9],"natural":[10],"processing,":[12],"enabling":[13],"diverse":[14,119],"applications":[15],"such":[16,98],"as":[17,99],"conversational":[18],"AI,":[19],"code":[20],"generation,":[21],"and":[22,32,80,124,143,160,183,205,223],"scientific":[23],"discovery.":[24],"Their":[25],"massive":[26],"parameter":[27],"counts":[28],"impose":[29],"significant":[30],"computational":[31],"memory":[33],"demands,":[34],"hindering":[35],"deployment":[36,96],"resource-constrained":[38],"environments.":[39],"Network":[40],"pruning":[41,59,75,108,145,181,188,197],"has":[42],"a":[44,73,104,137,169,218],"practical":[45],"solution":[46],"used":[47],"to":[48,83,92,117,153,179],"compress":[49,154],"LLMs":[50,155,207],"by":[51,149],"removing":[52],"their":[53],"redundant":[54],"parameters.":[55],"However,":[56],"existing":[57],"LLM":[58],"methods":[60],"are":[61],"primarily":[62],"designed":[63],"based":[64],"on":[65,121,203],"either":[66],"structured":[67,142],"or":[68],"unstructured":[69,144],"pruning,":[70],"typically":[71],"applying":[72],"uniform":[74],"strategy":[76],"across":[77,115,146],"all":[78],"layers,":[79],"they":[81],"fail":[82],"account":[84],"for":[85,107,199,225],"the":[86,195,211],"heterogeneous":[87,187],"sensitivity":[88,178],"of":[89,164],"different":[90,180],"layers":[91,116,147],"sparsification.":[93],"In":[94,127],"real-world":[95],"scenarios":[97],"edge-device":[100],"inference,":[101],"there":[102],"is":[103],"pressing":[105],"need":[106],"strategies":[109,114],"that":[110,139,174,190],"dynamically":[111],"adjust":[112],"sparsity":[113,213],"satisfy":[118],"requirements":[120],"latency,":[122],"accuracy,":[123],"energy":[125],"efficiency.":[126,161],"this":[128,192],"work,":[129],"we":[130],"propose":[131],"Layer-wise":[132],"Heterogeneity-guided":[133],"Heterogeneous":[134],"Pruning":[135],"(HHP),":[136],"method":[138,189,198,216],"adaptively":[140],"combines":[141],"guided":[148],"each":[150,176,200],"layer\u2019s":[151,177],"heterogeneity":[152,171],"while":[156],"balancing":[157],"inference":[158],"performance":[159,222],"HHP":[162],"consists":[163],"two":[165],"key":[166],"components:":[167],"(1)":[168],"layer-wise":[170],"importance":[172],"score":[173,193],"quantifies":[175],"granularities,":[182],"(2)":[184],"an":[185],"adaptive":[186],"uses":[191],"select":[194],"appropriate":[196],"layer.":[201],"Experiments":[202],"LLaMA":[204],"other":[206],"show":[208],"that,":[209],"at":[210],"40%":[212],"ratio,":[214],"our":[215],"provides":[217],"controllable":[219],"trade-off":[220],"between":[221],"efficiency":[224],"pruned":[226],"models.":[227]},"counts_by_year":[],"updated_date":"2026-08-21T09:56:20.448147","created_date":"2026-05-05T00:00:00"}