{"id":"https://openalex.org/W4389520768","doi":"https://doi.org/10.18653/v1/2023.emnlp-main.565","title":"EasyQuant: An Efficient Data-free Quantization Algorithm for LLMs","display_name":"EasyQuant: An Efficient Data-free Quantization Algorithm for LLMs","publication_year":2023,"publication_date":"2023-01-01","ids":{"openalex":"https://openalex.org/W4389520768","doi":"https://doi.org/10.18653/v1/2023.emnlp-main.565"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2023.emnlp-main.565","doi":"https://doi.org/10.18653/v1/2023.emnlp-main.565","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.emnlp-main.565","pdf_url":"https://aclanthology.org/2023.emnlp-main.565.pdf","source":{"id":"https://openalex.org/S4363608991","display_name":"Conference on Empirical Methods in Natural Language Processing (EMNLP)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","is_traditional":true,"indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://aclanthology.org/2023.emnlp-main.565.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5089829380","display_name":"Hanlin Tang","orcid":"https://orcid.org/0000-0002-6613-9174","observed_orcids":["https://orcid.org/0000-0002-6613-9174","https://orcid.org/0000-0003-0306-1637"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hanlin Tang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5075553304","display_name":"Yifu Sun","orcid":"https://orcid.org/0000-0003-4924-9387","observed_orcids":["https://orcid.org/0000-0003-4924-9387"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yifu Sun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5110593952","display_name":"Decheng Wu","orcid":"https://orcid.org/0000-0001-7317-1634","observed_orcids":["https://orcid.org/0000-0001-7317-1634"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Decheng Wu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101970899","display_name":"Kai Liu","orcid":"https://orcid.org/0000-0003-4529-3356","observed_orcids":["https://orcid.org/0000-0003-4529-3356"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kai Liu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5059894156","display_name":"Jianchen Zhu","orcid":"https://orcid.org/0000-0002-5988-3704","observed_orcids":["https://orcid.org/0000-0002-5988-3704"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jianchen Zhu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5020128898","display_name":"Zhanhui Kang","orcid":"https://orcid.org/0009-0006-5151-4222","observed_orcids":["https://orcid.org/0009-0006-5151-4222"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhanhui Kang","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":1.4949,"has_fulltext":true,"cited_by_count":17,"citation_normalized_percentile":{"value":0.89416143,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":95,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"9119","last_page":"9128"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9775000214576721,"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"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9775000214576721,"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"}},{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.0038999998942017555,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.00279999990016222,"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/data-free-quantization","display_name":"data-free quantization","score":0.859000027179718},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.8389999866485596},{"id":"https://openalex.org/keywords/weight-only-quantization","display_name":"weight-only quantization","score":0.9929999709129333},{"id":"https://openalex.org/keywords/reconstruction-error","display_name":"reconstruction error","score":0.765999972820282},{"id":"https://openalex.org/keywords/quantization-error","display_name":"quantization error","score":0.5400000214576721},{"id":"https://openalex.org/keywords/model-compression","display_name":"model compression","score":0.6159999966621399}],"concepts":[{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.8561221361160278},{"id":"https://openalex.org/C93372532","wikidata":"https://www.wikidata.org/wiki/Q6552455","display_name":"Linde\u2013Buzo\u2013Gray algorithm","level":3,"score":0.6102613806724548},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6031457185745239},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.5596531629562378},{"id":"https://openalex.org/C79337645","wikidata":"https://www.wikidata.org/wiki/Q779824","display_name":"Outlier","level":2,"score":0.5005218982696533},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2112804353237152}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2023.emnlp-main.565","doi":"https://doi.org/10.18653/v1/2023.emnlp-main.565","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.emnlp-main.565","pdf_url":"https://aclanthology.org/2023.emnlp-main.565.pdf","source":{"id":"https://openalex.org/S4363608991","display_name":"Conference on Empirical Methods in Natural Language Processing (EMNLP)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2403.02775","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2403.02775","pdf_url":"https://arxiv.org/pdf/2403.02775","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":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2403.02775","doi":"https://doi.org/10.48550/arxiv.2403.02775","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2403.02775","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.18653/v1/2023.emnlp-main.565","doi":"https://doi.org/10.18653/v1/2023.emnlp-main.565","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.emnlp-main.565","pdf_url":"https://aclanthology.org/2023.emnlp-main.565.pdf","source":{"id":"https://openalex.org/S4363608991","display_name":"Conference on Empirical Methods in Natural Language Processing (EMNLP)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing","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/W4389520768.pdf","grobid_xml":"https://content.openalex.org/works/W4389520768.grobid-xml"},"referenced_works_count":29,"referenced_works":["https://openalex.org/W1674493795","https://openalex.org/W2198190323","https://openalex.org/W2267635276","https://openalex.org/W2525332836","https://openalex.org/W2626778328","https://openalex.org/W2692059227","https://openalex.org/W2777406049","https://openalex.org/W2806507179","https://openalex.org/W2896457183","https://openalex.org/W2979314664","https://openalex.org/W2981852735","https://openalex.org/W2981910001","https://openalex.org/W3012638462","https://openalex.org/W3030163527","https://openalex.org/W3044934451","https://openalex.org/W3098576111","https://openalex.org/W3127829048","https://openalex.org/W4221167110","https://openalex.org/W4224308101","https://openalex.org/W4226146865","https://openalex.org/W4229005866","https://openalex.org/W4303443398","https://openalex.org/W4307934016","https://openalex.org/W4309591680","https://openalex.org/W4318751318","https://openalex.org/W4322718191","https://openalex.org/W7133198810","https://openalex.org/W7133204437","https://openalex.org/W7133231030"],"related_works":["https://openalex.org/W7222959735","https://openalex.org/W4378770729","https://openalex.org/W4399912076","https://openalex.org/W4399554365","https://openalex.org/W4309591680","https://openalex.org/W4400703137","https://openalex.org/W4404988798","https://openalex.org/W4386231419","https://openalex.org/W4400104239","https://openalex.org/W4386080707"],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"(LLMs)":[3],"have":[4],"proven":[5],"to":[6,10,65,87,140,155],"be":[7,178,188],"very":[8],"superior":[9],"conventional":[11],"methods":[12],"in":[13,39,70,112,125,180,190],"various":[14],"tasks.However,":[15],"their":[16],"expensive":[17],"computations":[18],"and":[19,68,99,115,135,218],"high":[20],"memory":[21],"requirements":[22],"are":[23,118,173,203],"prohibitive":[24],"for":[25,32,85,104,120,195],"deployment.Model":[26],"quantization":[27,83,102,116,123,138],"is":[28,37],"an":[29,75],"effective":[30],"method":[31,84],"reducing":[33,121],"this":[34,71,92],"overhead.The":[35],"problem":[36],"that":[38,108,150,183,207],"most":[40],"previous":[41],"works,":[42],"the":[43,53,59,62,113,122,129,137,142,156,167,184,204,227],"quantized":[44,63,171,185],"model":[45,186],"was":[46],"calibrated":[47],"using":[48],"a":[49,81,97,191,215],"few":[50,192],"samples":[51],"from":[52],"training":[54,165],"data,":[55,166],"which":[56],"might":[57],"affect":[58],"generalization":[60,90,168],"of":[61,170],"LLMs":[64,86,172,196],"unknown":[66],"cases":[67],"tasks.Hence":[69],"work,":[72,93],"we":[73,79,94,127,147,202],"explore":[74],"important":[76],"question:":[77],"Can":[78],"design":[80],"data-free":[82,100,216],"guarantee":[88],"its":[89],"performance?In":[91],"propose":[95],"EasyQuant,":[96,126],"trainingfree":[98],"weight-only":[101],"algorithm":[103,220],"LLMs.Our":[105],"observation":[106],"indicates":[107],"two":[109],"factors:":[110],"outliers":[111,130],"weight":[114],"ranges,":[117],"essential":[119],"error.Therefore,":[124],"leave":[128],"(less":[131],"than":[132,226],"1%)":[133],"unchanged":[134],"optimize":[136],"range":[139],"reduce":[141],"reconstruction":[143],"error.With":[144],"these":[145],"methods,":[146],"surprisingly":[148],"find":[149],"EasyQuant":[151,159,176],"achieves":[152,208],"comparable":[153,209],"performance":[154,169,210],"original":[157],"model.Since":[158],"does":[160],"not":[161],"depend":[162],"on":[163],"any":[164],"safely":[174],"guaranteed.Moreover,":[175],"can":[177],"implemented":[179],"parallel":[181],"so":[182],"could":[187],"attained":[189],"minutes":[193],"even":[194],"over":[197,222],"100B.To":[198],"our":[199,219],"best":[200],"knowledge,":[201],"first":[205],"work":[206],"with":[211],"datadependent":[212],"algorithms":[213],"under":[214],"setting":[217],"runs":[221],"10":[223],"times":[224],"faster":[225],"data-dependent":[228],"methods.":[229]},"counts_by_year":[{"year":2026,"cited_by_count":8},{"year":2025,"cited_by_count":6},{"year":2024,"cited_by_count":3}],"updated_date":"2026-11-08T12:12:28.789561","created_date":"2023-12-11T00:00:00"}