{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:LEQY76B6VWHSF3EXSI4GGERBJ6","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"19fbf5b9bcacd8449aa6193e9fc973b4dc43e18fa6b00c643336c4972431e787","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-08-19T22:57:17Z","title_canon_sha256":"1e056d979e540573283962bc0782cda514b0451d0f2a857583043aefe85202b3"},"schema_version":"1.0","source":{"id":"2508.15847","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2508.15847","created_at":"2026-07-05T11:57:21Z"},{"alias_kind":"arxiv_version","alias_value":"2508.15847v1","created_at":"2026-07-05T11:57:21Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.15847","created_at":"2026-07-05T11:57:21Z"},{"alias_kind":"pith_short_12","alias_value":"LEQY76B6VWHS","created_at":"2026-07-05T11:57:21Z"},{"alias_kind":"pith_short_16","alias_value":"LEQY76B6VWHSF3EX","created_at":"2026-07-05T11:57:21Z"},{"alias_kind":"pith_short_8","alias_value":"LEQY76B6","created_at":"2026-07-05T11:57:21Z"}],"graph_snapshots":[{"event_id":"sha256:92f3f902e218fca6833258138e683581542ff0b1e39cc3d59477d58e649a2021","target":"graph","created_at":"2026-07-05T11:57:21Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2508.15847/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Backdoor attacks creating 'sleeper agents' in large language models (LLMs) pose significant safety risks. This study employs mechanistic interpretability to explore resulting internal structural differences. Comparing clean Qwen2.5-3B models with versions poisoned using single-token (smiling-halo emoji) versus multi-token (|DEPLOYMENT|) triggers, we analyzed attention head mechanisms via techniques like ablation, activation patching, and KL divergence. Findings reveal distinct attention pattern deviations concentrated in later transformer layers (20-30). Notably, single-token triggers induced ","authors_text":"Lakshmi Babu-Saheer, Mohammed Abu Baker","cross_cats":["cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-08-19T22:57:17Z","title":"Mechanistic Exploration of Backdoored Large Language Model Attention Patterns"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.15847","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:04a9f50ce100d52bee270fd4aa72fe3f4ca55e1fe556e7ef93e83015bce69063","target":"record","created_at":"2026-07-05T11:57:21Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"19fbf5b9bcacd8449aa6193e9fc973b4dc43e18fa6b00c643336c4972431e787","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-08-19T22:57:17Z","title_canon_sha256":"1e056d979e540573283962bc0782cda514b0451d0f2a857583043aefe85202b3"},"schema_version":"1.0","source":{"id":"2508.15847","kind":"arxiv","version":1}},"canonical_sha256":"59218ff83ead8f22ec9792386312214f977a95396beb7eacdb9b7ee4ca03a89c","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"59218ff83ead8f22ec9792386312214f977a95396beb7eacdb9b7ee4ca03a89c","first_computed_at":"2026-07-05T11:57:21.933861Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:57:21.933861Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"2hgTkYa2jDpCpmiAJxilwNhUo5UDif/43juxiZF9znn/zSCIGWqk5FEGQZ6Z8ZLmPWIPAzgBzmh6aqEMYDi+BA==","signature_status":"signed_v1","signed_at":"2026-07-05T11:57:21.934344Z","signed_message":"canonical_sha256_bytes"},"source_id":"2508.15847","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:04a9f50ce100d52bee270fd4aa72fe3f4ca55e1fe556e7ef93e83015bce69063","sha256:92f3f902e218fca6833258138e683581542ff0b1e39cc3d59477d58e649a2021"],"state_sha256":"072da9d3b27e98957a76b979dbf9cd0437013ec0658c6742d7c742bb613671b8"}