{"id":"https://openalex.org/W4414744725","doi":"https://doi.org/10.52202/085713-1938","title":"DISCO: Disentangled Communication Steering for Large Language Models","display_name":"DISCO: Disentangled Communication Steering for Large Language Models","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4414744725","doi":"https://doi.org/10.52202/085713-1938"},"language":"en","primary_location":{"id":"doi:10.52202/085713-1938","doi":"https://doi.org/10.52202/085713-1938","is_oa":false,"landing_page_url":"https://doi.org/10.52202/085713-1938","pdf_url":null,"source":{"id":"https://openalex.org/S4363606243","display_name":"neural information processing systems","issn_l":"1049-5258","issn":["1049-5258"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","norway-2"],"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":"Advances in Neural Information Processing Systems 38","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","is_traditional":true,"indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2509.16820","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5020197423","display_name":"Max Torop","orcid":"https://orcid.org/0000-0001-7873-108X","observed_orcids":["https://orcid.org/0000-0001-7873-108X"]},"institutions":[{"id":"https://openalex.org/I12912129","display_name":"Northeastern University","ror":"https://ror.org/04t5xt781","country_code":"US","type":"education","lineage":["https://openalex.org/I12912129"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Max Torop","raw_affiliation_strings":["Northeastern University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Northeastern University","institution_ids":["https://openalex.org/I12912129"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5011474053","display_name":"Aria Masoomi","orcid":"https://orcid.org/0009-0005-5822-1716","observed_orcids":["https://orcid.org/0009-0005-5822-1716"]},"institutions":[{"id":"https://openalex.org/I12912129","display_name":"Northeastern University","ror":"https://ror.org/04t5xt781","country_code":"US","type":"education","lineage":["https://openalex.org/I12912129"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Aria Masoomi","raw_affiliation_strings":["Northeastern University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Northeastern University","institution_ids":["https://openalex.org/I12912129"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5040839375","display_name":"Masih Eskandar","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I12912129","display_name":"Northeastern University","ror":"https://ror.org/04t5xt781","country_code":"US","type":"education","lineage":["https://openalex.org/I12912129"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Masih Eskandar","raw_affiliation_strings":["Northeastern University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Northeastern University","institution_ids":["https://openalex.org/I12912129"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5042038501","display_name":"Jennifer Dy","orcid":"https://orcid.org/0000-0002-8430-134X","observed_orcids":["https://orcid.org/0000-0002-8430-134X"]},"institutions":[{"id":"https://openalex.org/I12912129","display_name":"Northeastern University","ror":"https://ror.org/04t5xt781","country_code":"US","type":"education","lineage":["https://openalex.org/I12912129"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jennifer Dy","raw_affiliation_strings":["Northeastern University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Northeastern University","institution_ids":["https://openalex.org/I12912129"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I12912129"],"apc_list":null,"apc_paid":null,"fwci":0.368,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":{"value":0.64253316,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":91,"max":95},"biblio":{"volume":null,"issue":null,"first_page":"64453","last_page":"64505"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.8637999892234802,"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":0.8637999892234802,"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.04349999874830246,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.02850000001490116,"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/steering-vectors","display_name":"steering vectors","score":0.9279999732971191},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.8889999985694885},{"id":"https://openalex.org/keywords/attention-heads","display_name":"attention heads","score":0.9290000200271606},{"id":"https://openalex.org/keywords/llama-3-1","display_name":"Llama 3.1","score":0.8600000143051147},{"id":"https://openalex.org/keywords/gemma-2","display_name":"Gemma 2","score":0.949999988079071}],"concepts":[{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.6832000017166138},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6765000224113464},{"id":"https://openalex.org/C136197465","wikidata":"https://www.wikidata.org/wiki/Q1729295","display_name":"Variety (cybernetics)","level":2,"score":0.6294999718666077},{"id":"https://openalex.org/C189950617","wikidata":"https://www.wikidata.org/wiki/Q937228","display_name":"Property (philosophy)","level":2,"score":0.5181000232696533},{"id":"https://openalex.org/C2776291640","wikidata":"https://www.wikidata.org/wiki/Q2912517","display_name":"Value (mathematics)","level":2,"score":0.4512999951839447},{"id":"https://openalex.org/C2780312720","wikidata":"https://www.wikidata.org/wiki/Q5689100","display_name":"Head (geology)","level":2,"score":0.4246000051498413},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.38760000467300415},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3801000118255615},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3474000096321106},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.323199987411499},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.30559998750686646},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.29429998993873596},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.28610000014305115},{"id":"https://openalex.org/C48677424","wikidata":"https://www.wikidata.org/wiki/Q6888088","display_name":"Mode (computer interface)","level":2,"score":0.2542000114917755},{"id":"https://openalex.org/C123079801","wikidata":"https://www.wikidata.org/wiki/Q750240","display_name":"Modulation (music)","level":2,"score":0.25380000472068787}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.52202/085713-1938","doi":"https://doi.org/10.52202/085713-1938","is_oa":false,"landing_page_url":"https://doi.org/10.52202/085713-1938","pdf_url":null,"source":{"id":"https://openalex.org/S4363606243","display_name":"neural information processing systems","issn_l":"1049-5258","issn":["1049-5258"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","norway-2"],"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":"Advances in Neural Information Processing Systems 38","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2509.16820","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2509.16820","pdf_url":"https://arxiv.org/pdf/2509.16820","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2509.16820","doi":"https://doi.org/10.48550/arxiv.2509.16820","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2509.16820","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2509.16820","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2509.16820","pdf_url":"https://arxiv.org/pdf/2509.16820","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4414744725.pdf","grobid_xml":"https://content.openalex.org/works/W4414744725.grobid-xml"},"referenced_works_count":16,"referenced_works":["https://openalex.org/W3037932933","https://openalex.org/W4285149215","https://openalex.org/W4327810158","https://openalex.org/W4387355345","https://openalex.org/W4387929831","https://openalex.org/W4390041933","https://openalex.org/W4400104375","https://openalex.org/W4400601054","https://openalex.org/W4401306886","https://openalex.org/W4401306979","https://openalex.org/W4402671231","https://openalex.org/W4404330874","https://openalex.org/W4406961517","https://openalex.org/W4406975150","https://openalex.org/W4407208939","https://openalex.org/W7133236469"],"related_works":["https://openalex.org/W7153319398","https://openalex.org/W7212078628","https://openalex.org/W7212035274","https://openalex.org/W7160637976","https://openalex.org/W4415036317","https://openalex.org/W7127443069","https://openalex.org/W7151634578","https://openalex.org/W7160597879","https://openalex.org/W4414801980","https://openalex.org/W7166827879"],"abstract_inverted_index":{"A":[0],"variety":[1],"of":[2,14,48,55,63,75,118,135],"recent":[3],"methods":[4],"guide":[5],"large":[6],"language":[7],"model":[8],"outputs":[9],"via":[10],"the":[11,32,61,73,159,164,167],"inference-time":[12],"addition":[13],"steering":[15,28,64,100,136,151,177],"vectors":[16,29],"to":[17,26,155],"residual-stream":[18],"or":[19],"attention-head":[20],"representations.":[21],"In":[22,113],"contrast,":[23,114],"we":[24,79],"propose":[25],"inject":[27],"directly":[30],"into":[31],"query":[33,168],"and":[34,107,146,169],"value":[35,170],"representation":[36],"spaces":[37,50,171],"within":[38],"attention":[39,67,86,101],"heads.":[40],"We":[41,70,125],"provide":[42],"evidence":[43],"that":[44,92,127,166],"a":[45,95,110,133],"greater":[46],"portion":[47],"these":[49,119],"exhibit":[51],"high":[52],"linear":[53],"discriminability":[54],"concepts":[56],"--a":[57],"key":[58],"property":[59],"motivating":[60],"use":[62],"vectors--":[65],"than":[66,158],"head":[68,87],"outputs.":[69,88],"analytically":[71],"characterize":[72],"effect":[74],"our":[76],"method,":[77],"which":[78,103],"term":[80],"DISentangled":[81],"COmmunication":[82],"(DISCO)":[83],"Steering,":[84],"on":[85,142],"Our":[89,161],"analysis":[90],"reveals":[91],"DISCO":[93,128],"disentangles":[94],"strong":[96],"but":[97],"underutilized":[98],"baseline,":[99],"inputs,":[102],"implicitly":[104],"modifies":[105],"queries":[106],"values":[108],"in":[109],"rigid":[111],"manner.":[112],"DISCO's":[115],"direct":[116],"modulation":[117],"components":[120],"enables":[121],"more":[122],"granular":[123],"control.":[124],"find":[126],"achieves":[129],"superior":[130],"performance":[131],"over":[132],"number":[134],"vector":[137,178],"baselines":[138],"across":[139],"multiple":[140],"datasets":[141],"LLaMA":[143],"3.1":[144],"8B":[145],"Gemma":[147],"2":[148],"9B,":[149],"with":[150],"efficacy":[152],"scoring":[153],"up":[154],"19.1%":[156],"higher":[157],"runner-up.":[160],"results":[162],"support":[163],"conclusion":[165],"are":[172],"powerful":[173],"building":[174],"blocks":[175],"for":[176],"methods.":[179]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2025-10-10T00:00:00"}