{"id":"https://openalex.org/W4396717865","doi":"https://doi.org/10.48550/arxiv.2405.02750","title":"Enhancing Contextual Understanding in Large Language Models through Contrastive Decoding","display_name":"Enhancing Contextual Understanding in Large Language Models through Contrastive Decoding","publication_year":2024,"publication_date":"2024-05-04","ids":{"openalex":"https://openalex.org/W4396717865","doi":"https://doi.org/10.48550/arxiv.2405.02750"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2405.02750","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2405.02750","pdf_url":"https://arxiv.org/pdf/2405.02750","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/2405.02750","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5007396467","display_name":"Zheng Zhao","orcid":"https://orcid.org/0000-0001-9586-268X"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Zhao, Zheng","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5021745136","display_name":"Emilio Monti","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Monti, Emilio","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5067133778","display_name":"Jens Lehmann","orcid":"https://orcid.org/0000-0001-9108-4278"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lehmann, Jens","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5101558340","display_name":"Haytham Assem","orcid":"https://orcid.org/0000-0001-6026-9683"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Assem, Haytham","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":4,"corresponding_author_ids":["https://openalex.org/A5007396467"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":0,"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":0.9617000222206116,"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":0.9617000222206116,"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":0.9455999732017517,"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/decoding-methods","display_name":"Decoding methods","score":0.7182023525238037},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.5803731679916382},{"id":"https://openalex.org/keywords/linguistics","display_name":"Linguistics","score":0.4841635823249817},{"id":"https://openalex.org/keywords/psychology","display_name":"Psychology","score":0.41245758533477783},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.3897828757762909},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.06922325491905212},{"id":"https://openalex.org/keywords/philosophy","display_name":"Philosophy","score":0.0638653039932251}],"concepts":[{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.7182023525238037},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5803731679916382},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.4841635823249817},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.41245758533477783},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3897828757762909},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.06922325491905212},{"id":"https://openalex.org/C138885662","wikidata":"https://www.wikidata.org/wiki/Q5891","display_name":"Philosophy","level":0,"score":0.0638653039932251}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2405.02750","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2405.02750","pdf_url":"https://arxiv.org/pdf/2405.02750","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.2405.02750","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2405.02750","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:2405.02750","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2405.02750","pdf_url":"https://arxiv.org/pdf/2405.02750","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":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4396717865.pdf","grobid_xml":"https://content.openalex.org/works/W4396717865.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052","https://openalex.org/W2382290278","https://openalex.org/W4395014643"],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"(LLMs)":[3],"tend":[4],"to":[5,99,121],"inadequately":[6],"integrate":[7],"input":[8,52],"context":[9,75,102],"during":[10,68,104],"text":[11,26],"generation,":[12],"relying":[13],"excessively":[14],"on":[15],"encoded":[16],"prior":[17,41],"knowledge":[18,38,43,50,66],"in":[19,24,73],"model":[20],"parameters,":[21],"potentially":[22],"resulting":[23],"generated":[25],"with":[27,92],"factual":[28],"inconsistencies":[29],"or":[30],"contextually":[31],"unfaithful":[32],"content.":[33],"LLMs":[34,62],"utilize":[35],"two":[36],"primary":[37],"sources:":[39],"1)":[40],"(parametric)":[42],"from":[44,51],"pretraining,":[45],"and":[46,125],"2)":[47],"contextual":[48],"(non-parametric)":[49],"prompts.":[53],"The":[54],"study":[55],"addresses":[56],"the":[57,69,74],"open":[58],"question":[59,78],"of":[60,76],"how":[61],"effectively":[63],"balance":[64],"these":[65],"sources":[67],"generation":[70],"process,":[71],"specifically":[72],"open-domain":[77],"answering.":[79],"To":[80],"address":[81],"this":[82],"issue,":[83],"we":[84],"introduce":[85],"a":[86],"novel":[87],"approach":[88],"integrating":[89],"contrastive":[90],"decoding":[91],"adversarial":[93],"irrelevant":[94],"passages":[95],"as":[96],"negative":[97],"samples":[98],"enhance":[100],"robust":[101],"grounding":[103],"generation.":[105],"Notably,":[106],"our":[107],"method":[108],"operates":[109],"at":[110],"inference":[111],"time":[112],"without":[113],"requiring":[114],"further":[115],"training.":[116],"We":[117],"conduct":[118],"comprehensive":[119],"experiments":[120],"demonstrate":[122],"its":[123,131],"applicability":[124],"effectiveness,":[126],"providing":[127],"empirical":[128],"evidence":[129],"showcasing":[130],"superiority":[132],"over":[133],"existing":[134],"methodologies.":[135],"Our":[136],"code":[137],"is":[138],"publicly":[139],"available":[140],"at:":[141],"https://github.com/amazon-science/ContextualUnderstanding-ContrastiveDecoding.":[142]},"counts_by_year":[],"updated_date":"2026-03-10T16:38:18.471706","created_date":"2025-10-10T00:00:00"}