{"id":"https://openalex.org/W4417536753","doi":"https://doi.org/10.48550/arxiv.2506.15735","title":"ContextBench: Modifying Contexts for Targeted Latent Activation","display_name":"ContextBench: Modifying Contexts for Targeted Latent Activation","publication_year":2025,"publication_date":"2025-06-15","ids":{"openalex":"https://openalex.org/W4417536753","doi":"https://doi.org/10.48550/arxiv.2506.15735"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2506.15735","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.15735","pdf_url":"https://arxiv.org/pdf/2506.15735","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"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2506.15735","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Graham, Robert","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Graham, Robert","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Stevinson, Edward","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I47508984","display_name":"Imperial College London","ror":"https://ror.org/041kmwe10","country_code":"GB","type":"education","lineage":["https://openalex.org/I47508984"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Stevinson, Edward","raw_affiliation_strings":["Imperial College London ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Imperial College London ,","institution_ids":["https://openalex.org/I47508984"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Richter, Leo","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I45129253","display_name":"University College London","ror":"https://ror.org/02jx3x895","country_code":"GB","type":"education","lineage":["https://openalex.org/I124357947","https://openalex.org/I45129253"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Richter, Leo","raw_affiliation_strings":["University College London"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University College London","institution_ids":["https://openalex.org/I45129253"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Chia, Alexander","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chia, Alexander","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Miller, Joseph","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I40120149","display_name":"University of Oxford","ror":"https://ror.org/052gg0110","country_code":"GB","type":"education","lineage":["https://openalex.org/I40120149"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Miller, Joseph","raw_affiliation_strings":["University of Oxford ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Oxford ,","institution_ids":["https://openalex.org/I40120149"]}]},{"author_position":"last","author":{"id":null,"display_name":"Bloom, Joseph Isaac","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bloom, Joseph Isaac","raw_affiliation_strings":["UK AI Security Institute"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"UK AI Security Institute","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"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":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.9878000020980835,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.9878000020980835,"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/T10028","display_name":"Topic Modeling","score":0.007199999876320362,"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":0.001500000013038516,"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/latent-features","display_name":"latent features","score":0.6000000238418579},{"id":"https://openalex.org/keywords/language-model-safety","display_name":"language model safety","score":0.8119999766349792},{"id":"https://openalex.org/keywords/benchmark","display_name":"benchmark","score":0.46799999475479126},{"id":"https://openalex.org/keywords/language-fluency","display_name":"language fluency","score":0.8569999933242798}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6647999882698059},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.6248000264167786},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.5871999859809875},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.48089998960494995},{"id":"https://openalex.org/C2777212361","wikidata":"https://www.wikidata.org/wiki/Q5127848","display_name":"Class (philosophy)","level":2,"score":0.4738999903202057},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.43689998984336853},{"id":"https://openalex.org/C2164484","wikidata":"https://www.wikidata.org/wiki/Q5170150","display_name":"Core (optical fiber)","level":2,"score":0.3955000042915344},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.39100000262260437},{"id":"https://openalex.org/C183322885","wikidata":"https://www.wikidata.org/wiki/Q17007702","display_name":"Context model","level":3,"score":0.3174000084400177},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3046000003814697},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.2994000017642975},{"id":"https://openalex.org/C51167844","wikidata":"https://www.wikidata.org/wiki/Q4422623","display_name":"Latent variable","level":2,"score":0.2976999878883362},{"id":"https://openalex.org/C168031717","wikidata":"https://www.wikidata.org/wiki/Q1530280","display_name":"Balance (ability)","level":2,"score":0.29649999737739563},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.29409998655319214},{"id":"https://openalex.org/C70727504","wikidata":"https://www.wikidata.org/wiki/Q1806878","display_name":"Latent class model","level":2,"score":0.2913999855518341},{"id":"https://openalex.org/C2777868144","wikidata":"https://www.wikidata.org/wiki/Q7239817","display_name":"Preference elicitation","level":3,"score":0.2793000042438507},{"id":"https://openalex.org/C2986053828","wikidata":"https://www.wikidata.org/wiki/Q355217","display_name":"Time budget","level":2,"score":0.25440001487731934}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2506.15735","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.15735","pdf_url":"https://arxiv.org/pdf/2506.15735","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.2506.15735","doi":"https://doi.org/10.48550/arxiv.2506.15735","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2506.15735","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":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2506.15735","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.15735","pdf_url":"https://arxiv.org/pdf/2506.15735","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":[],"sustainable_development_goals_aurora":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4417536753.pdf","grobid_xml":"https://content.openalex.org/works/W4417536753.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Identifying":[0],"inputs":[1,33],"that":[2,34,106],"trigger":[3],"specific":[4,36],"behaviours":[5],"or":[6,39,77],"latent":[7,37,75],"features":[8,38,76],"in":[9,112],"language":[10],"models":[11],"could":[12],"have":[13],"a":[14,23,54],"wide":[15],"range":[16],"of":[17,25,28,74],"safety":[18,64],"use":[19],"cases.":[20],"We":[21,43,92],"investigate":[22],"class":[24],"methods":[26,86],"capable":[27],"generating":[29],"targeted,":[30],"linguistically":[31],"fluent":[32],"activate":[35],"elicit":[40],"model":[41,102],"behaviours.":[42],"formalise":[44],"this":[45],"approach":[46],"as":[47],"context":[48],"modification":[49],"and":[50,62,79,100,104,116],"present":[51],"ContextBench":[52],"--":[53],"benchmark":[55],"with":[56,98],"tasks":[57],"assessing":[58],"core":[59],"method":[60],"capabilities":[61],"potential":[63],"applications.":[65],"Our":[66],"evaluation":[67],"framework":[68],"measures":[69],"both":[70],"elicitation":[71,114],"strength":[72],"(activation":[73],"behaviours)":[78],"linguistic":[80],"fluency,":[81],"highlighting":[82],"how":[83],"current":[84],"state-of-the-art":[85,110],"struggle":[87],"to":[88],"balance":[89],"these":[90,107],"objectives.":[91],"enhance":[93],"Evolutionary":[94],"Prompt":[95],"Optimisation":[96],"(EPO)":[97],"LLM-assistance":[99],"diffusion":[101],"inpainting,":[103],"demonstrate":[105],"variants":[108],"achieve":[109],"performance":[111],"balancing":[113],"effectiveness":[115],"fluency.":[117]},"counts_by_year":[],"updated_date":"2026-09-30T09:27:27.532063","created_date":"2025-10-10T00:00:00"}