{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:ZM3BKB2JAC3HJLUOM5FUS3OMTP","short_pith_number":"pith:ZM3BKB2J","schema_version":"1.0","canonical_sha256":"cb3615074900b674ae8e674b496dcc9bf47853a2b78cb27568cb02a1ed8fab57","source":{"kind":"arxiv","id":"2606.03780","version":1},"attestation_state":"computed","paper":{"title":"Expert-Aware Causal Tracing of Factual Recall in Sparse MoE Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Ali Modarressi, Hinrich Sch\\\"utze, Yihong Liu, Yuetian Lu","submitted_at":"2026-06-02T15:35:48Z","abstract_excerpt":"Causal tracing of factual recall has been studied predominantly in dense transformer language models, where interventions localize information flow to layers or feed-forward modules. Sparse mixture-of-experts (MoE) language models introduce a sharper question: when a factual prediction is mediated by a routed MoE block, which routed expert contributions matter? We formulate expert-aware causal tracing for sparse MoE language models. Using CounterFact facts, we first corrupt the model's factual preference by adding noise to subject-token embeddings, and then test whether clean MoE-block outputs"},"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":"2606.03780","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-06-02T15:35:48Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"2a2ff1e97826612171b5df9f3e78b84caf4e530b50ac96339a3d7a18e1f59bbc","abstract_canon_sha256":"62405713bb170ba3cb73cbc9563d4123df37758330fa4e7e129f098ea90442b3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-03T02:06:02.242886Z","signature_b64":"DxXsPAdoSxK28HNFG5/jIIu3Tj7OkcpER/7ggMDBlvjs7+VhmjLD0/nug1oS66Apq0l3ULZEGqn93Wl6AdMjAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cb3615074900b674ae8e674b496dcc9bf47853a2b78cb27568cb02a1ed8fab57","last_reissued_at":"2026-06-03T02:06:02.242469Z","signature_status":"signed_v1","first_computed_at":"2026-06-03T02:06:02.242469Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Expert-Aware Causal Tracing of Factual Recall in Sparse MoE Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Ali Modarressi, Hinrich Sch\\\"utze, Yihong Liu, Yuetian Lu","submitted_at":"2026-06-02T15:35:48Z","abstract_excerpt":"Causal tracing of factual recall has been studied predominantly in dense transformer language models, where interventions localize information flow to layers or feed-forward modules. Sparse mixture-of-experts (MoE) language models introduce a sharper question: when a factual prediction is mediated by a routed MoE block, which routed expert contributions matter? We formulate expert-aware causal tracing for sparse MoE language models. Using CounterFact facts, we first corrupt the model's factual preference by adding noise to subject-token embeddings, and then test whether clean MoE-block outputs"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.03780","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/2606.03780/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":"2606.03780","created_at":"2026-06-03T02:06:02.242533+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.03780v1","created_at":"2026-06-03T02:06:02.242533+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.03780","created_at":"2026-06-03T02:06:02.242533+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZM3BKB2JAC3H","created_at":"2026-06-03T02:06:02.242533+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZM3BKB2JAC3HJLUO","created_at":"2026-06-03T02:06:02.242533+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZM3BKB2J","created_at":"2026-06-03T02:06:02.242533+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/ZM3BKB2JAC3HJLUOM5FUS3OMTP","json":"https://pith.science/pith/ZM3BKB2JAC3HJLUOM5FUS3OMTP.json","graph_json":"https://pith.science/api/pith-number/ZM3BKB2JAC3HJLUOM5FUS3OMTP/graph.json","events_json":"https://pith.science/api/pith-number/ZM3BKB2JAC3HJLUOM5FUS3OMTP/events.json","paper":"https://pith.science/paper/ZM3BKB2J"},"agent_actions":{"view_html":"https://pith.science/pith/ZM3BKB2JAC3HJLUOM5FUS3OMTP","download_json":"https://pith.science/pith/ZM3BKB2JAC3HJLUOM5FUS3OMTP.json","view_paper":"https://pith.science/paper/ZM3BKB2J","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.03780&json=true","fetch_graph":"https://pith.science/api/pith-number/ZM3BKB2JAC3HJLUOM5FUS3OMTP/graph.json","fetch_events":"https://pith.science/api/pith-number/ZM3BKB2JAC3HJLUOM5FUS3OMTP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZM3BKB2JAC3HJLUOM5FUS3OMTP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZM3BKB2JAC3HJLUOM5FUS3OMTP/action/storage_attestation","attest_author":"https://pith.science/pith/ZM3BKB2JAC3HJLUOM5FUS3OMTP/action/author_attestation","sign_citation":"https://pith.science/pith/ZM3BKB2JAC3HJLUOM5FUS3OMTP/action/citation_signature","submit_replication":"https://pith.science/pith/ZM3BKB2JAC3HJLUOM5FUS3OMTP/action/replication_record"}},"created_at":"2026-06-03T02:06:02.242533+00:00","updated_at":"2026-06-03T02:06:02.242533+00:00"}