{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:IKJMA223JJOMC7VB3FLR5545YS","short_pith_number":"pith:IKJMA223","schema_version":"1.0","canonical_sha256":"4292c06b5b4a5cc17ea1d9571ef79dc4b36f7fca5840fbdf2906739200d92c7f","source":{"kind":"arxiv","id":"2505.19333","version":1},"attestation_state":"computed","paper":{"title":"Evaluating Steering Techniques using Human Similarity Judgments","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Kushin Mukherjee, Siddharth Suresh, Timothy T. Rogers, Zach Studdiford","submitted_at":"2025-05-25T21:40:26Z","abstract_excerpt":"Current evaluations of Large Language Model (LLM) steering techniques focus on task-specific performance, overlooking how well steered representations align with human cognition. Using a well-established triadic similarity judgment task, we assessed steered LLMs on their ability to flexibly judge similarity between concepts based on size or kind. We found that prompt-based steering methods outperformed other methods both in terms of steering accuracy and model-to-human alignment. We also found LLMs were biased towards 'kind' similarity and struggled with 'size' alignment. This evaluation appro"},"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":"2505.19333","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-05-25T21:40:26Z","cross_cats_sorted":[],"title_canon_sha256":"d46774c8fa21c102bae28605f6f628f9ff2ad43bc3dae12284977746796b3ea2","abstract_canon_sha256":"9ee341dd972a34f211f69dc6546f0c8469b3ea7c452925cdb73abc6b4d04cc6b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:09:28.535471Z","signature_b64":"cHoMRgOWfTy5qfbCR3faT1+4jAJnCOE34xKWCSM7vRKCBz4G7gtYhffce5X2wiCGvau2jX/aZd/cGkjaAZxvDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4292c06b5b4a5cc17ea1d9571ef79dc4b36f7fca5840fbdf2906739200d92c7f","last_reissued_at":"2026-07-05T11:09:28.534913Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:09:28.534913Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Evaluating Steering Techniques using Human Similarity Judgments","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Kushin Mukherjee, Siddharth Suresh, Timothy T. Rogers, Zach Studdiford","submitted_at":"2025-05-25T21:40:26Z","abstract_excerpt":"Current evaluations of Large Language Model (LLM) steering techniques focus on task-specific performance, overlooking how well steered representations align with human cognition. Using a well-established triadic similarity judgment task, we assessed steered LLMs on their ability to flexibly judge similarity between concepts based on size or kind. We found that prompt-based steering methods outperformed other methods both in terms of steering accuracy and model-to-human alignment. We also found LLMs were biased towards 'kind' similarity and struggled with 'size' alignment. This evaluation appro"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.19333","kind":"arxiv","version":1},"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/2505.19333/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":"2505.19333","created_at":"2026-07-05T11:09:28.534980+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.19333v1","created_at":"2026-07-05T11:09:28.534980+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.19333","created_at":"2026-07-05T11:09:28.534980+00:00"},{"alias_kind":"pith_short_12","alias_value":"IKJMA223JJOM","created_at":"2026-07-05T11:09:28.534980+00:00"},{"alias_kind":"pith_short_16","alias_value":"IKJMA223JJOMC7VB","created_at":"2026-07-05T11:09:28.534980+00:00"},{"alias_kind":"pith_short_8","alias_value":"IKJMA223","created_at":"2026-07-05T11:09:28.534980+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/IKJMA223JJOMC7VB3FLR5545YS","json":"https://pith.science/pith/IKJMA223JJOMC7VB3FLR5545YS.json","graph_json":"https://pith.science/api/pith-number/IKJMA223JJOMC7VB3FLR5545YS/graph.json","events_json":"https://pith.science/api/pith-number/IKJMA223JJOMC7VB3FLR5545YS/events.json","paper":"https://pith.science/paper/IKJMA223"},"agent_actions":{"view_html":"https://pith.science/pith/IKJMA223JJOMC7VB3FLR5545YS","download_json":"https://pith.science/pith/IKJMA223JJOMC7VB3FLR5545YS.json","view_paper":"https://pith.science/paper/IKJMA223","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.19333&json=true","fetch_graph":"https://pith.science/api/pith-number/IKJMA223JJOMC7VB3FLR5545YS/graph.json","fetch_events":"https://pith.science/api/pith-number/IKJMA223JJOMC7VB3FLR5545YS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IKJMA223JJOMC7VB3FLR5545YS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IKJMA223JJOMC7VB3FLR5545YS/action/storage_attestation","attest_author":"https://pith.science/pith/IKJMA223JJOMC7VB3FLR5545YS/action/author_attestation","sign_citation":"https://pith.science/pith/IKJMA223JJOMC7VB3FLR5545YS/action/citation_signature","submit_replication":"https://pith.science/pith/IKJMA223JJOMC7VB3FLR5545YS/action/replication_record"}},"created_at":"2026-07-05T11:09:28.534980+00:00","updated_at":"2026-07-05T11:09:28.534980+00:00"}