{"id":"https://openalex.org/W4412945404","doi":"https://doi.org/10.18653/v1/2025.acl-long.506","title":"Towards Context-Robust LLMs: A Gated Representation Fine-tuning Approach","display_name":"Towards Context-Robust LLMs: A Gated Representation Fine-tuning Approach","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412945404","doi":"https://doi.org/10.18653/v1/2025.acl-long.506"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.acl-long.506","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.506","pdf_url":"https://aclanthology.org/2025.acl-long.506.pdf","source":null,"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 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.acl-long.506.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5111125512","display_name":"Shenglai Zeng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shenglai Zeng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101533342","display_name":"Peng He","orcid":"https://orcid.org/0000-0002-0788-9872"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Pengfei He","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103182012","display_name":"Kai Guo","orcid":"https://orcid.org/0000-0002-7831-2122"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kai Guo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5053505580","display_name":"Tianqi Zheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tianqi Zheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102146587","display_name":"Hanqing Lu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hanqing Lu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5056505716","display_name":"Yue Xing","orcid":"https://orcid.org/0009-0001-2499-7461"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yue Xing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5018073436","display_name":"Hui Liu","orcid":"https://orcid.org/0000-0003-4599-5702"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hui Liu","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":null,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"10262","last_page":"10276"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9832000136375427,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9832000136375427,"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/context","display_name":"Context (archaeology)","score":0.6273788809776306},{"id":"https://openalex.org/keywords/representation","display_name":"Representation (politics)","score":0.6139323115348816},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.5709176063537598},{"id":"https://openalex.org/keywords/political-science","display_name":"Political science","score":0.14869007468223572},{"id":"https://openalex.org/keywords/history","display_name":"History","score":0.11532208323478699}],"concepts":[{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.6273788809776306},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.6139323115348816},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5709176063537598},{"id":"https://openalex.org/C17744445","wikidata":"https://www.wikidata.org/wiki/Q36442","display_name":"Political science","level":0,"score":0.14869007468223572},{"id":"https://openalex.org/C95457728","wikidata":"https://www.wikidata.org/wiki/Q309","display_name":"History","level":0,"score":0.11532208323478699},{"id":"https://openalex.org/C199539241","wikidata":"https://www.wikidata.org/wiki/Q7748","display_name":"Law","level":1,"score":0.0},{"id":"https://openalex.org/C94625758","wikidata":"https://www.wikidata.org/wiki/Q7163","display_name":"Politics","level":2,"score":0.0},{"id":"https://openalex.org/C166957645","wikidata":"https://www.wikidata.org/wiki/Q23498","display_name":"Archaeology","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.acl-long.506","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.506","pdf_url":"https://aclanthology.org/2025.acl-long.506.pdf","source":null,"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 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.acl-long.506","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.506","pdf_url":"https://aclanthology.org/2025.acl-long.506.pdf","source":null,"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 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G5808857907","display_name":null,"funder_award_id":"DRL2405483","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G647087074","display_name":null,"funder_award_id":"IOS2107215","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G7402679956","display_name":null,"funder_award_id":"IOS2035472","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G7794432752","display_name":null,"funder_award_id":"IIS2212144","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G8083687766","display_name":null,"funder_award_id":"CNS2246050","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320306111","display_name":"U.S. Department of Commerce","ror":"https://ror.org/04chq2495"},{"id":"https://openalex.org/F4320308150","display_name":"Michigan Department of Agriculture and Rural Development","ror":"https://ror.org/023cd6641"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412945404.pdf","grobid_xml":"https://content.openalex.org/works/W4412945404.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052"],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"enhanced":[4],"with":[5,50,122],"external":[6,25,51,63,75],"contexts,":[7],"such":[8],"as":[9],"through":[10],"retrieval-augmented":[11],"generation":[12],"(RAG),":[13],"often":[14],"face":[15],"challenges":[16],"in":[17],"handling":[18],"imperfect":[19],"evidence.They":[20],"tend":[21],"to":[22,30,54,103,113],"over-rely":[23],"on":[24,62,128],"knowledge,":[26,69,76],"making":[27],"them":[28],"vulnerable":[29],"misleading":[31],"and":[32,74,77,89,105,109],"unhelpful":[33,79],"contexts.To":[34,80],"address":[35],"this,":[36],"we":[37,84],"propose":[38],"the":[39],"concept":[40],"of":[41,96,125],"context-robust":[42,58,139],"LLMs,":[43],"which":[44],"can":[45,134],"effectively":[46,135],"balance":[47],"internal":[48,68,73],"knowledge":[49],"context,":[52],"similar":[53],"human":[55],"cognitive":[56],"processes.Specifically,":[57],"LLMs":[59,137],"should":[60],"rely":[61],"context":[64],"only":[65,123],"when":[66],"lacking":[67],"identify":[70],"contradictions":[71],"between":[72],"disregard":[78],"achieve":[81],"this":[82],"goal,":[83],"introduce":[85],"Grft,":[86],"a":[87,100,118],"lightweight":[88,119],"plug-and-play":[90],"gated":[91],"representation":[92,111],"fine-tuning":[93],"approach.Grft":[94],"consists":[95],"two":[97],"key":[98],"components:":[99],"gating":[101],"mechanism":[102],"detect":[104],"filter":[106],"problematic":[107],"inputs,":[108],"low-rank":[110],"adapters":[112],"adjust":[114],"hidden":[115],"representations.By":[116],"training":[117],"intervention":[120],"function":[121],"0.0004%":[124],"model":[126],"size":[127],"fewer":[129],"than":[130],"200":[131],"examples,":[132],"Grft":[133],"adapts":[136],"towards":[138],"behaviors.":[140]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}