{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:VJROJVCAP5IJ4DP7FSKK3V7TWG","short_pith_number":"pith:VJROJVCA","schema_version":"1.0","canonical_sha256":"aa62e4d4407f509e0dff2c94add7f3b1976ebb7d25e2a15f187fd5467d38521f","source":{"kind":"arxiv","id":"2606.09899","version":1},"attestation_state":"computed","paper":{"title":"When Attribution Patching Lies: Diagnosis and a Second-Order Correction","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Jialu Wang, Luyang Zhang","submitted_at":"2026-06-05T05:20:32Z","abstract_excerpt":"A central goal of mechanistic interpretability is to identify which internal components causally drive a language model's behavior. Because these importance estimates serve as the evidence for identifying circuits, systematic errors can lead to the misidentification of the underlying mechanisms. While activation patching provides a gold-standard causal metric, its computational cost is prohibitive at scale. Practitioners instead rely on attribution patching, a gradient-based, first-order approximation whose reliability remains poorly understood. In this work, we characterize the source of this"},"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":"2606.09899","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-05T05:20:32Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"d489cfee78c064a58b81bc79d692ea47eaa40a80a8c91055d20724528d142adb","abstract_canon_sha256":"1da4dc6dbfeede94edf8e456d6f83444bb59450e15d07e378e4cd705fc686d93"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-10T00:08:32.069686Z","signature_b64":"XGObt2Y+ev/dT05T+9vEJr6sbmKO5Eh/dPP0fokONg7E+i0eQBA4VZGoRHKnONR9qQDHnb9nBqqR56WlsGU6CA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"aa62e4d4407f509e0dff2c94add7f3b1976ebb7d25e2a15f187fd5467d38521f","last_reissued_at":"2026-06-10T00:08:32.068903Z","signature_status":"signed_v1","first_computed_at":"2026-06-10T00:08:32.068903Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"When Attribution Patching Lies: Diagnosis and a Second-Order Correction","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Jialu Wang, Luyang Zhang","submitted_at":"2026-06-05T05:20:32Z","abstract_excerpt":"A central goal of mechanistic interpretability is to identify which internal components causally drive a language model's behavior. Because these importance estimates serve as the evidence for identifying circuits, systematic errors can lead to the misidentification of the underlying mechanisms. While activation patching provides a gold-standard causal metric, its computational cost is prohibitive at scale. Practitioners instead rely on attribution patching, a gradient-based, first-order approximation whose reliability remains poorly understood. In this work, we characterize the source of this"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.09899","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/2606.09899/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":"2606.09899","created_at":"2026-06-10T00:08:32.069028+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.09899v1","created_at":"2026-06-10T00:08:32.069028+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.09899","created_at":"2026-06-10T00:08:32.069028+00:00"},{"alias_kind":"pith_short_12","alias_value":"VJROJVCAP5IJ","created_at":"2026-06-10T00:08:32.069028+00:00"},{"alias_kind":"pith_short_16","alias_value":"VJROJVCAP5IJ4DP7","created_at":"2026-06-10T00:08:32.069028+00:00"},{"alias_kind":"pith_short_8","alias_value":"VJROJVCA","created_at":"2026-06-10T00:08:32.069028+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/VJROJVCAP5IJ4DP7FSKK3V7TWG","json":"https://pith.science/pith/VJROJVCAP5IJ4DP7FSKK3V7TWG.json","graph_json":"https://pith.science/api/pith-number/VJROJVCAP5IJ4DP7FSKK3V7TWG/graph.json","events_json":"https://pith.science/api/pith-number/VJROJVCAP5IJ4DP7FSKK3V7TWG/events.json","paper":"https://pith.science/paper/VJROJVCA"},"agent_actions":{"view_html":"https://pith.science/pith/VJROJVCAP5IJ4DP7FSKK3V7TWG","download_json":"https://pith.science/pith/VJROJVCAP5IJ4DP7FSKK3V7TWG.json","view_paper":"https://pith.science/paper/VJROJVCA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.09899&json=true","fetch_graph":"https://pith.science/api/pith-number/VJROJVCAP5IJ4DP7FSKK3V7TWG/graph.json","fetch_events":"https://pith.science/api/pith-number/VJROJVCAP5IJ4DP7FSKK3V7TWG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VJROJVCAP5IJ4DP7FSKK3V7TWG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VJROJVCAP5IJ4DP7FSKK3V7TWG/action/storage_attestation","attest_author":"https://pith.science/pith/VJROJVCAP5IJ4DP7FSKK3V7TWG/action/author_attestation","sign_citation":"https://pith.science/pith/VJROJVCAP5IJ4DP7FSKK3V7TWG/action/citation_signature","submit_replication":"https://pith.science/pith/VJROJVCAP5IJ4DP7FSKK3V7TWG/action/replication_record"}},"created_at":"2026-06-10T00:08:32.069028+00:00","updated_at":"2026-06-10T00:08:32.069028+00:00"}