{"id":"https://openalex.org/W4392619028","doi":"https://doi.org/10.48550/arxiv.2403.03853","title":"ShortGPT: Layers in Large Language Models are More Redundant Than You Expect","display_name":"ShortGPT: Layers in Large Language Models are More Redundant Than You Expect","publication_year":2024,"publication_date":"2024-03-06","ids":{"openalex":"https://openalex.org/W4392619028","doi":"https://doi.org/10.48550/arxiv.2403.03853"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2403.03853","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2403.03853","pdf_url":"https://arxiv.org/pdf/2403.03853","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"text"},"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2403.03853","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5109724260","display_name":"Xin Men","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Men, Xin","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5018995861","display_name":"Mingyu Xu","orcid":"https://orcid.org/0009-0007-6347-0551"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Mingyu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100681539","display_name":"Qingyu Zhang","orcid":"https://orcid.org/0000-0001-9739-3899"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Qingyu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5067986384","display_name":"Bingning Wang","orcid":"https://orcid.org/0009-0007-7748-7098"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Bingning","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101366403","display_name":"Hongyu Lin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lin, Hongyu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100585978","display_name":"Yaojie Lu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lu, Yaojie","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100620300","display_name":"Xianpei Han","orcid":"https://orcid.org/0000-0002-1304-6302"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Han, Xianpei","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5056336684","display_name":"Weipeng Chen","orcid":"https://orcid.org/0000-0001-9293-7578"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Weipeng","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":8,"corresponding_author_ids":["https://openalex.org/A5109724260"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":15,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9516000151634216,"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":0.9516000151634216,"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":0.904699981212616,"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":0.5002539157867432},{"id":"https://openalex.org/keywords/linguistics","display_name":"Linguistics","score":0.39518389105796814},{"id":"https://openalex.org/keywords/philosophy","display_name":"Philosophy","score":0.15123224258422852}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5002539157867432},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.39518389105796814},{"id":"https://openalex.org/C138885662","wikidata":"https://www.wikidata.org/wiki/Q5891","display_name":"Philosophy","level":0,"score":0.15123224258422852}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2403.03853","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2403.03853","pdf_url":"https://arxiv.org/pdf/2403.03853","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"text"},{"id":"doi:10.48550/arxiv.2403.03853","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2403.03853","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2403.03853","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2403.03853","pdf_url":"https://arxiv.org/pdf/2403.03853","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"text"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4392619028.pdf"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W2358668433","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W2382290278","https://openalex.org/W2478288626","https://openalex.org/W4391913857","https://openalex.org/W2350741829","https://openalex.org/W2530322880"],"abstract_inverted_index":{"As":[0],"Large":[1],"Language":[2],"Models":[3],"(LLMs)":[4],"continue":[5],"to":[6,61,116,128,138],"advance":[7],"in":[8,26,46,68,79,87,109,122,149],"performance,":[9],"their":[10,91],"size":[11],"has":[12],"escalated":[13],"significantly,":[14],"with":[15],"current":[16],"LLMs":[17,35,88],"containing":[18],"billions":[19],"or":[20],"even":[21],"trillions":[22],"of":[23,34,65,147],"parameters.":[24],"However,":[25],"this":[27,51],"study,":[28],"we":[29,53,81,100],"discovered":[30],"that":[31,96],"many":[32],"layers":[33,41,86],"exhibit":[36],"high":[37,145],"similarity,":[38],"and":[39,124],"some":[40],"play":[42],"a":[43,55,73,144],"negligible":[44],"role":[45],"network":[47],"functionality.":[48],"Based":[49],"on":[50,90],"observation,":[52],"define":[54],"metric":[56],"called":[57],"Block":[58],"Influence":[59],"(BI)":[60],"gauge":[62],"the":[63,84,150],"significance":[64],"each":[66],"layer":[67,77,134],"LLMs.":[69],"We":[70],"then":[71],"propose":[72],"straightforward":[74],"pruning":[75,141],"approach:":[76],"removal,":[78,135],"which":[80,99],"directly":[82],"delete":[83],"redundant":[85],"based":[89],"BI":[92],"scores.":[93],"Experiments":[94],"demonstrate":[95],"our":[97],"method,":[98],"call":[101],"ShortGPT,":[102],"significantly":[103],"outperforms":[104],"previous":[105],"state-of-the-art":[106],"(SOTA)":[107],"methods":[108],"model":[110,151],"pruning.":[111],"Moreover,":[112],"ShortGPT":[113],"is":[114],"orthogonal":[115],"quantization-like":[117],"methods,":[118],"enabling":[119],"further":[120],"reduction":[121],"parameters":[123],"computation.":[125],"The":[126],"ability":[127],"achieve":[129],"better":[130],"results":[131],"through":[132],"simple":[133],"as":[136],"opposed":[137],"more":[139],"complex":[140],"techniques,":[142],"suggests":[143],"degree":[146],"redundancy":[148],"architecture.":[152]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":5},{"year":2024,"cited_by_count":9}],"updated_date":"2026-03-24T08:02:53.985720","created_date":"2025-10-10T00:00:00"}