{"id":"https://openalex.org/W7124947693","doi":"https://doi.org/10.26615/978-954-452-098-4-136","title":"LLM Compression: How Far CanWe Go in Balancing Size and Performance?","display_name":"LLM Compression: How Far CanWe Go in Balancing Size and Performance?","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W7124947693","doi":"https://doi.org/10.26615/978-954-452-098-4-136"},"language":"en","primary_location":{"id":"doi:10.26615/978-954-452-098-4-136","doi":"https://doi.org/10.26615/978-954-452-098-4-136","is_oa":true,"landing_page_url":"https://doi.org/10.26615/978-954-452-098-4-136","pdf_url":"https://doi.org/10.26615/978-954-452-098-4-136","source":{"id":"https://openalex.org/S4393917233","display_name":"International conference Recent advances in natural language processing","issn_l":"1313-8502","issn":["1313-8502","2603-2813"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":["jufo-1","norway-1"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"International Conference Recent Advances in Natural Language Processing","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","is_traditional":true,"indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"bronze","oa_url":"https://doi.org/10.26615/978-954-452-098-4-136","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5123435871","display_name":"India Odia Generative AI","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Odia Generative AI, India","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5123416777","display_name":"Sahil Sk","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sahil Sk","raw_affiliation_strings":["Odia Generative AI , India;"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Odia Generative AI , India;","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5093557689","display_name":"Debasish Dhal","orcid":"https://orcid.org/0009-0009-2217-8831","observed_orcids":["https://orcid.org/0009-0009-2217-8831"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Debasish Dhal","raw_affiliation_strings":["Odia Generative AI , India;"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Odia Generative AI , India;","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5123435871","display_name":"India Odia Generative AI","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Odia Generative AI, India","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5049576827","display_name":"Sonal Khosla","orcid":"https://orcid.org/0000-0002-9325-1639","observed_orcids":["https://orcid.org/0000-0002-9325-1639"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sonal Khosla","raw_affiliation_strings":["Odia Generative AI , India;"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Odia Generative AI , India;","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5123435871","display_name":"India Odia Generative AI","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Odia Generative AI, India","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103289470","display_name":"SK Shahid","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sk Shahid","raw_affiliation_strings":["Odia Generative AI , India;"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Odia Generative AI , India;","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5123435871","display_name":"India Odia Generative AI","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Odia Generative AI, India","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102942268","display_name":"Sambit Sekhar","orcid":"https://orcid.org/0009-0005-4698-8004","observed_orcids":["https://orcid.org/0009-0005-4698-8004"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sambit Shekhar","raw_affiliation_strings":["Odia Generative AI , India;"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Odia Generative AI , India;","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5123435871","display_name":"India Odia Generative AI","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Odia Generative AI, India","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5018479257","display_name":"Akash Kumar Dhaka","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Akash Dhaka","raw_affiliation_strings":["AMD Silo AI , Finland"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"AMD Silo AI , Finland","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5123416835","display_name":"Finland AMD Silo AI","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"AMD Silo AI, Finland","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5015497545","display_name":"Shantipriya Parida","orcid":"https://orcid.org/0000-0003-3387-6300","observed_orcids":["https://orcid.org/0000-0003-3387-6300"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shantipriya Parida","raw_affiliation_strings":["AMD Silo AI , Finland"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"AMD Silo AI , Finland","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5123416835","display_name":"Finland AMD Silo AI","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"AMD Silo AI, Finland","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5064348917","display_name":"Dilip Kumar Prasad","orcid":"https://orcid.org/0000-0002-3693-6973","observed_orcids":["https://orcid.org/0000-0002-3693-6973"]},"institutions":[{"id":"https://openalex.org/I78037679","display_name":"UiT The Arctic University of Norway","ror":"https://ror.org/00wge5k78","country_code":"NO","type":"education","lineage":["https://openalex.org/I78037679"]}],"countries":["NO"],"is_corresponding":false,"raw_author_name":"Dilip K. Prasad","raw_affiliation_strings":["The Arctic University of Norway , Norway"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Arctic University of Norway , Norway","institution_ids":["https://openalex.org/I78037679"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5077746237","display_name":"Ond\u0159ej Bojar","orcid":"https://orcid.org/0000-0002-0606-0050","observed_orcids":["https://orcid.org/0000-0002-0606-0050"]},"institutions":[{"id":"https://openalex.org/I21250087","display_name":"Charles University","ror":"https://ror.org/024d6js02","country_code":"CZ","type":"education","lineage":["https://openalex.org/I21250087"]}],"countries":["CZ"],"is_corresponding":false,"raw_author_name":"Ond\u0159ej Bojar","raw_affiliation_strings":["Charles University , MFF , \u00daFAL , Czech Republic"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Charles University , MFF , \u00daFAL , Czech Republic","institution_ids":["https://openalex.org/I21250087"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.0,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"2025","issue":null,"first_page":"1183","last_page":"1187"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.8452000021934509,"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":0.8452000021934509,"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":0.11020000278949738,"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":0.009399999864399433,"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/large-language-model-compression","display_name":"large language model compression","score":0.9160000085830688},{"id":"https://openalex.org/keywords/low-bit-quantization","display_name":"low-bit quantization","score":0.7910000085830688},{"id":"https://openalex.org/keywords/gptq","display_name":"GPTQ","score":0.968999981880188},{"id":"https://openalex.org/keywords/llama","display_name":"llamas","score":0.7979999780654907},{"id":"https://openalex.org/keywords/qwen","display_name":"Qwen","score":0.9789999723434448},{"id":"https://openalex.org/keywords/phi","display_name":"phi","score":0.6949999928474426},{"id":"https://openalex.org/keywords/inference-latency","display_name":"inference latency","score":0.6269999742507935},{"id":"https://openalex.org/keywords/throughput","display_name":"throughput","score":0.9559999704360962},{"id":"https://openalex.org/keywords/information-retrieval","display_name":"information retrieval","score":0.7509999871253967},{"id":"https://openalex.org/keywords/mathematical-reasoning","display_name":"mathematical reasoning","score":0.9860000014305115}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4422999918460846},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.25369998812675476},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.24009999632835388},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.23819999396800995},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.23080000281333923},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.22789999842643738},{"id":"https://openalex.org/C9652623","wikidata":"https://www.wikidata.org/wiki/Q190109","display_name":"Field (mathematics)","level":2,"score":0.2257000058889389},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.22540000081062317},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.21770000457763672},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.21379999816417694}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.26615/978-954-452-098-4-136","doi":"https://doi.org/10.26615/978-954-452-098-4-136","is_oa":true,"landing_page_url":"https://doi.org/10.26615/978-954-452-098-4-136","pdf_url":"https://doi.org/10.26615/978-954-452-098-4-136","source":{"id":"https://openalex.org/S4393917233","display_name":"International conference Recent advances in natural language processing","issn_l":"1313-8502","issn":["1313-8502","2603-2813"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":["jufo-1","norway-1"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"International Conference Recent Advances in Natural Language Processing","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2508.11318","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2508.11318","pdf_url":"https://arxiv.org/pdf/2508.11318","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"}],"best_oa_location":{"id":"doi:10.26615/978-954-452-098-4-136","doi":"https://doi.org/10.26615/978-954-452-098-4-136","is_oa":true,"landing_page_url":"https://doi.org/10.26615/978-954-452-098-4-136","pdf_url":"https://doi.org/10.26615/978-954-452-098-4-136","source":{"id":"https://openalex.org/S4393917233","display_name":"International conference Recent advances in natural language processing","issn_l":"1313-8502","issn":["1313-8502","2603-2813"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":["jufo-1","norway-1"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"International Conference Recent Advances in Natural Language Processing","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G8155346842","display_name":null,"funder_award_id":"325741","funder_id":"https://openalex.org/F4320323299","funder_display_name":"Norges Forskningsr\u00e5d"}],"funders":[{"id":"https://openalex.org/F4320323299","display_name":"Norges Forskningsr\u00e5d","ror":"https://ror.org/00epmv149"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7124947693.pdf","grobid_xml":"https://content.openalex.org/works/W7124947693.grobid-xml"},"referenced_works_count":6,"referenced_works":["https://openalex.org/W2951534261","https://openalex.org/W2953212265","https://openalex.org/W3210904070","https://openalex.org/W4307934016","https://openalex.org/W4386080707","https://openalex.org/W4396913356"],"related_works":["https://openalex.org/W6986681499","https://openalex.org/W7161755584","https://openalex.org/W4396881553","https://openalex.org/W7143343316","https://openalex.org/W7155032781","https://openalex.org/W7168082978","https://openalex.org/W7204622377","https://openalex.org/W7162429959","https://openalex.org/W4415310986","https://openalex.org/W4392340571"],"abstract_inverted_index":{"Quantization":[0,33,39],"is":[1],"an":[2],"essential":[3],"and":[4,20,35,46,68,76,88,99,140,144],"popular":[5],"technique":[6],"for":[7,115,158],"improving":[8],"the":[9,83,110,118,130,138],"accessibility":[10],"of":[11,112,142,149],"large":[12],"language":[13],"models":[14,58,148],"(LLMs)":[15],"by":[16],"reducing":[17],"memory":[18],"usage":[19],"computational":[21],"costs":[22],"while":[23],"maintaining":[24],"performance.In":[25],"this":[26],"study,":[27],"we":[28],"apply":[29],"4-bit":[30],"Group":[31],"Scaling":[32],"(GSQ)":[34],"Generative":[36],"Pretrained":[37],"Transformer":[38],"(GPTQ)":[40],"to":[41,134],"LLaMA":[42],"1B,":[43],"Qwen":[44],"0.5B,":[45],"PHI":[47],"1.5B,":[48],"evaluating":[49],"their":[50],"impact":[51],"across":[52,78],"multiple":[53],"NLP":[54],"tasks.We":[55],"benchmark":[56,157],"these":[57],"on":[59,129,147],"MS":[60],"MARCO":[61],"(Information":[62],"Retrieval),":[63],"BoolQ":[64],"(Boolean":[65],"Question":[66],"Answering),":[67],"GSM8K":[69],"(Mathematical":[70],"Reasoning)":[71],"datasets,":[72],"assessing":[73],"both":[74],"accuracy":[75],"efficiency":[77],"various":[79],"tasks.The":[80],"study":[81],"measures":[82],"tradeoffs":[84],"between":[85],"model":[86],"compression":[87],"task":[89],"performance,":[90],"analyzing":[91],"key":[92],"evaluation":[93],"metrics":[94],"namely:":[95],"accuracy,":[96],"inference":[97],"latency,":[98],"throughput":[100],"(total":[101],"output":[102],"tokens":[103],"generated":[104],"per":[105],"second),":[106],"providing":[107],"insights":[108],"into":[109],"suitability":[111],"low-bit":[113],"quantization":[114],"real-world":[116],"deployment.Using":[117],"results,":[119],"a":[120,125,156],"user":[121],"can":[122],"then":[123],"make":[124],"suitable":[126],"decision":[127],"based":[128],"specifications":[131],"that":[132],"need":[133],"be":[135],"met.We":[136],"discuss":[137],"pros":[139],"cons":[141],"GSQ":[143],"GPTQ":[145],"techniques":[146],"different":[150],"sizes,":[151],"which":[152],"also":[153],"serve":[154],"as":[155],"future":[159],"experiments.":[160]},"counts_by_year":[],"updated_date":"2026-11-04T10:54:05.112534","created_date":"2026-01-21T00:00:00"}