{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:VAUXFR5AJ4AWQSC4EMCCNBEYCB","short_pith_number":"pith:VAUXFR5A","schema_version":"1.0","canonical_sha256":"a82972c7a04f0168485c23042684981074de80a3575fbc6c1f0434d28ad3f12c","source":{"kind":"arxiv","id":"2407.18712","version":2},"attestation_state":"computed","paper":{"title":"Cluster-norm for Unsupervised Probing of Knowledge","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Gr\\'egoire Dhimo\\\"ila, Kaarel H\\\"anni, Owen (Ho Wan) Yeung, Sharan Maiya, Walter Laurito","submitted_at":"2024-07-26T12:57:54Z","abstract_excerpt":"The deployment of language models brings challenges in generating reliable information, especially when these models are fine-tuned using human preferences. To extract encoded knowledge without (potentially) biased human labels, unsupervised probing techniques like Contrast-Consistent Search (CCS) have been developed (Burns et al., 2022). However, salient but unrelated features in a given dataset can mislead these probes (Farquhar et al., 2023). Addressing this, we propose a cluster normalization method to minimize the impact of such features by clustering and normalizing activations of contra"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2407.18712","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2024-07-26T12:57:54Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"132e7235f577d4c72cab1903cd6a6ec5a7bb9e2a87810cf253194897069fbad2","abstract_canon_sha256":"ef4a7363820103f440af1355cd254641b27ad3173aa788f7e86a5de388ca6771"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:15:35.312836Z","signature_b64":"ddgj86xWGiK6k4t7kQZUZBcLJAbQ6H4WUdzEn8sqJwMd0yeI0OGTCpWCSeVxnvTGRYmyxXoox0eusnIg0N6/Ag==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a82972c7a04f0168485c23042684981074de80a3575fbc6c1f0434d28ad3f12c","last_reissued_at":"2026-07-05T09:15:35.312304Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:15:35.312304Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Cluster-norm for Unsupervised Probing of Knowledge","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Gr\\'egoire Dhimo\\\"ila, Kaarel H\\\"anni, Owen (Ho Wan) Yeung, Sharan Maiya, Walter Laurito","submitted_at":"2024-07-26T12:57:54Z","abstract_excerpt":"The deployment of language models brings challenges in generating reliable information, especially when these models are fine-tuned using human preferences. To extract encoded knowledge without (potentially) biased human labels, unsupervised probing techniques like Contrast-Consistent Search (CCS) have been developed (Burns et al., 2022). However, salient but unrelated features in a given dataset can mislead these probes (Farquhar et al., 2023). Addressing this, we propose a cluster normalization method to minimize the impact of such features by clustering and normalizing activations of contra"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.18712","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2407.18712/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2407.18712","created_at":"2026-07-05T09:15:35.312383+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.18712v2","created_at":"2026-07-05T09:15:35.312383+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.18712","created_at":"2026-07-05T09:15:35.312383+00:00"},{"alias_kind":"pith_short_12","alias_value":"VAUXFR5AJ4AW","created_at":"2026-07-05T09:15:35.312383+00:00"},{"alias_kind":"pith_short_16","alias_value":"VAUXFR5AJ4AWQSC4","created_at":"2026-07-05T09:15:35.312383+00:00"},{"alias_kind":"pith_short_8","alias_value":"VAUXFR5A","created_at":"2026-07-05T09:15:35.312383+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VAUXFR5AJ4AWQSC4EMCCNBEYCB","json":"https://pith.science/pith/VAUXFR5AJ4AWQSC4EMCCNBEYCB.json","graph_json":"https://pith.science/api/pith-number/VAUXFR5AJ4AWQSC4EMCCNBEYCB/graph.json","events_json":"https://pith.science/api/pith-number/VAUXFR5AJ4AWQSC4EMCCNBEYCB/events.json","paper":"https://pith.science/paper/VAUXFR5A"},"agent_actions":{"view_html":"https://pith.science/pith/VAUXFR5AJ4AWQSC4EMCCNBEYCB","download_json":"https://pith.science/pith/VAUXFR5AJ4AWQSC4EMCCNBEYCB.json","view_paper":"https://pith.science/paper/VAUXFR5A","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.18712&json=true","fetch_graph":"https://pith.science/api/pith-number/VAUXFR5AJ4AWQSC4EMCCNBEYCB/graph.json","fetch_events":"https://pith.science/api/pith-number/VAUXFR5AJ4AWQSC4EMCCNBEYCB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VAUXFR5AJ4AWQSC4EMCCNBEYCB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VAUXFR5AJ4AWQSC4EMCCNBEYCB/action/storage_attestation","attest_author":"https://pith.science/pith/VAUXFR5AJ4AWQSC4EMCCNBEYCB/action/author_attestation","sign_citation":"https://pith.science/pith/VAUXFR5AJ4AWQSC4EMCCNBEYCB/action/citation_signature","submit_replication":"https://pith.science/pith/VAUXFR5AJ4AWQSC4EMCCNBEYCB/action/replication_record"}},"created_at":"2026-07-05T09:15:35.312383+00:00","updated_at":"2026-07-05T09:15:35.312383+00:00"}