{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:74PTVZBMG4UGQLHPUZPNUTR6Y5","short_pith_number":"pith:74PTVZBM","schema_version":"1.0","canonical_sha256":"ff1f3ae42c3728682cefa65eda4e3ec7460dc9dce9ce1805ad7980e3e97d89bc","source":{"kind":"arxiv","id":"2510.05709","version":2},"attestation_state":"computed","paper":{"title":"Correcting Prompt Dependence in LLM Benchmarks: A Bayesian Hierarchical Model with Embedding-Space Clustering","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CR","authors_text":"Annie Gray, Isobel Thornton, James Bishop, Mary Llewellyn","submitted_at":"2025-10-07T09:22:22Z","abstract_excerpt":"LLM benchmarking metrics often misstate performance and uncertainty as they rely on two assumptions that frequently do not hold in practice: (i) a sufficient number of evaluations are available for classical inference, and (ii) test prompts are independent. We propose a corrective Bayesian hierarchical model with embedding-space clustering that provides robust performance metrics in limited-data settings while correcting for prompt dependence. We apply the approach to adversarial robustness benchmarks, showing consistent recovery of clustering structure, resulting in more reliable performance "},"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":"2510.05709","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CR","submitted_at":"2025-10-07T09:22:22Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"4e18b29d9f13a804e3375affd741e73cd68c186a7c9a39c801d0d03b6b16d288","abstract_canon_sha256":"cffdc3eb4bd170415122ef86e494fe508f316853c8a3500a2e3bd1c716a3e9e8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-05T01:15:16.842755Z","signature_b64":"c0H8v85ezNy8c21h1bgillB5NPrMoek0FKjxBDNvLuvoNDFjKuV3iez5NZG5RWdHTiIiGk7nt24JA/OAYGPQAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ff1f3ae42c3728682cefa65eda4e3ec7460dc9dce9ce1805ad7980e3e97d89bc","last_reissued_at":"2026-06-05T01:15:16.841830Z","signature_status":"signed_v1","first_computed_at":"2026-06-05T01:15:16.841830Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Correcting Prompt Dependence in LLM Benchmarks: A Bayesian Hierarchical Model with Embedding-Space Clustering","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CR","authors_text":"Annie Gray, Isobel Thornton, James Bishop, Mary Llewellyn","submitted_at":"2025-10-07T09:22:22Z","abstract_excerpt":"LLM benchmarking metrics often misstate performance and uncertainty as they rely on two assumptions that frequently do not hold in practice: (i) a sufficient number of evaluations are available for classical inference, and (ii) test prompts are independent. We propose a corrective Bayesian hierarchical model with embedding-space clustering that provides robust performance metrics in limited-data settings while correcting for prompt dependence. We apply the approach to adversarial robustness benchmarks, showing consistent recovery of clustering structure, resulting in more reliable performance "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.05709","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/2510.05709/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":"2510.05709","created_at":"2026-06-05T01:15:16.841962+00:00"},{"alias_kind":"arxiv_version","alias_value":"2510.05709v2","created_at":"2026-06-05T01:15:16.841962+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.05709","created_at":"2026-06-05T01:15:16.841962+00:00"},{"alias_kind":"pith_short_12","alias_value":"74PTVZBMG4UG","created_at":"2026-06-05T01:15:16.841962+00:00"},{"alias_kind":"pith_short_16","alias_value":"74PTVZBMG4UGQLHP","created_at":"2026-06-05T01:15:16.841962+00:00"},{"alias_kind":"pith_short_8","alias_value":"74PTVZBM","created_at":"2026-06-05T01:15:16.841962+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/74PTVZBMG4UGQLHPUZPNUTR6Y5","json":"https://pith.science/pith/74PTVZBMG4UGQLHPUZPNUTR6Y5.json","graph_json":"https://pith.science/api/pith-number/74PTVZBMG4UGQLHPUZPNUTR6Y5/graph.json","events_json":"https://pith.science/api/pith-number/74PTVZBMG4UGQLHPUZPNUTR6Y5/events.json","paper":"https://pith.science/paper/74PTVZBM"},"agent_actions":{"view_html":"https://pith.science/pith/74PTVZBMG4UGQLHPUZPNUTR6Y5","download_json":"https://pith.science/pith/74PTVZBMG4UGQLHPUZPNUTR6Y5.json","view_paper":"https://pith.science/paper/74PTVZBM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2510.05709&json=true","fetch_graph":"https://pith.science/api/pith-number/74PTVZBMG4UGQLHPUZPNUTR6Y5/graph.json","fetch_events":"https://pith.science/api/pith-number/74PTVZBMG4UGQLHPUZPNUTR6Y5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/74PTVZBMG4UGQLHPUZPNUTR6Y5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/74PTVZBMG4UGQLHPUZPNUTR6Y5/action/storage_attestation","attest_author":"https://pith.science/pith/74PTVZBMG4UGQLHPUZPNUTR6Y5/action/author_attestation","sign_citation":"https://pith.science/pith/74PTVZBMG4UGQLHPUZPNUTR6Y5/action/citation_signature","submit_replication":"https://pith.science/pith/74PTVZBMG4UGQLHPUZPNUTR6Y5/action/replication_record"}},"created_at":"2026-06-05T01:15:16.841962+00:00","updated_at":"2026-06-05T01:15:16.841962+00:00"}