{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:7OVMTO5BTGLZB43LLYULRCA2HL","short_pith_number":"pith:7OVMTO5B","schema_version":"1.0","canonical_sha256":"fbaac9bba1999790f36b5e28b8881a3ae81a8fc8b75fed28ac6f6c54e835be22","source":{"kind":"arxiv","id":"2502.10928","version":2},"attestation_state":"computed","paper":{"title":"Probing Semantic Routing in Large Mixture-of-Expert Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Chendi Xue, Man Luo, Matthew Lyle Olson, Musashi Hinck, Neale Ratzlaff, Sungduk Yu, Vasudev Lal","submitted_at":"2025-02-15T23:37:32Z","abstract_excerpt":"In the past year, large (>100B parameter) mixture-of-expert (MoE) models have become increasingly common in the open domain. While their advantages are often framed in terms of efficiency, prior work has also explored functional differentiation through routing behavior. We investigate whether expert routing in large MoE models is influenced by the semantics of the inputs. To test this, we design two controlled experiments. First, we compare activations on sentence pairs with a shared target word used in the same or different senses. Second, we fix context and substitute the target word with se"},"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":"2502.10928","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-15T23:37:32Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"511a5cf9030b1e122033d97e033deff0bfe45184017441769c9fbb4f36c3e7ad","abstract_canon_sha256":"249b655ac39618bf9c28e2d85033d65e8d4d2be712fab91508b781f53e20ee7d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:06:54.985580Z","signature_b64":"yv63KB/fcCt9K1/eliw9FpSk2O1B7/C/4qfwPPFIQqJBS+YSbydzf9m131deEwnSmpWOHDC4W68zr5cfqdHADA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fbaac9bba1999790f36b5e28b8881a3ae81a8fc8b75fed28ac6f6c54e835be22","last_reissued_at":"2026-07-05T11:06:54.985044Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:06:54.985044Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Probing Semantic Routing in Large Mixture-of-Expert Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Chendi Xue, Man Luo, Matthew Lyle Olson, Musashi Hinck, Neale Ratzlaff, Sungduk Yu, Vasudev Lal","submitted_at":"2025-02-15T23:37:32Z","abstract_excerpt":"In the past year, large (>100B parameter) mixture-of-expert (MoE) models have become increasingly common in the open domain. While their advantages are often framed in terms of efficiency, prior work has also explored functional differentiation through routing behavior. We investigate whether expert routing in large MoE models is influenced by the semantics of the inputs. To test this, we design two controlled experiments. First, we compare activations on sentence pairs with a shared target word used in the same or different senses. Second, we fix context and substitute the target word with se"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.10928","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/2502.10928/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":"2502.10928","created_at":"2026-07-05T11:06:54.985105+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.10928v2","created_at":"2026-07-05T11:06:54.985105+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.10928","created_at":"2026-07-05T11:06:54.985105+00:00"},{"alias_kind":"pith_short_12","alias_value":"7OVMTO5BTGLZ","created_at":"2026-07-05T11:06:54.985105+00:00"},{"alias_kind":"pith_short_16","alias_value":"7OVMTO5BTGLZB43L","created_at":"2026-07-05T11:06:54.985105+00:00"},{"alias_kind":"pith_short_8","alias_value":"7OVMTO5B","created_at":"2026-07-05T11:06:54.985105+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.04160","citing_title":"Expert-Aware Refusal Steering","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16930","citing_title":"CoGR-MoE: Concept-Guided Expert Routing with Consistent Selection and Flexible Reasoning for Visual Question Answering","ref_index":28,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7OVMTO5BTGLZB43LLYULRCA2HL","json":"https://pith.science/pith/7OVMTO5BTGLZB43LLYULRCA2HL.json","graph_json":"https://pith.science/api/pith-number/7OVMTO5BTGLZB43LLYULRCA2HL/graph.json","events_json":"https://pith.science/api/pith-number/7OVMTO5BTGLZB43LLYULRCA2HL/events.json","paper":"https://pith.science/paper/7OVMTO5B"},"agent_actions":{"view_html":"https://pith.science/pith/7OVMTO5BTGLZB43LLYULRCA2HL","download_json":"https://pith.science/pith/7OVMTO5BTGLZB43LLYULRCA2HL.json","view_paper":"https://pith.science/paper/7OVMTO5B","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.10928&json=true","fetch_graph":"https://pith.science/api/pith-number/7OVMTO5BTGLZB43LLYULRCA2HL/graph.json","fetch_events":"https://pith.science/api/pith-number/7OVMTO5BTGLZB43LLYULRCA2HL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7OVMTO5BTGLZB43LLYULRCA2HL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7OVMTO5BTGLZB43LLYULRCA2HL/action/storage_attestation","attest_author":"https://pith.science/pith/7OVMTO5BTGLZB43LLYULRCA2HL/action/author_attestation","sign_citation":"https://pith.science/pith/7OVMTO5BTGLZB43LLYULRCA2HL/action/citation_signature","submit_replication":"https://pith.science/pith/7OVMTO5BTGLZB43LLYULRCA2HL/action/replication_record"}},"created_at":"2026-07-05T11:06:54.985105+00:00","updated_at":"2026-07-05T11:06:54.985105+00:00"}