{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2I5ZCPBBPID5T6ZCGIB77LJXTM","short_pith_number":"pith:2I5ZCPBB","schema_version":"1.0","canonical_sha256":"d23b913c217a07d9fb223203ffad379b226e5548b3f44ce8e9cb8ed7217ae624","source":{"kind":"arxiv","id":"2504.04444","version":1},"attestation_state":"computed","paper":{"title":"On the Spatial Structure of Mixture-of-Experts in Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Daniel Bershatsky, Ivan Oseledets","submitted_at":"2025-04-06T11:31:55Z","abstract_excerpt":"A common assumption is that MoE routers primarily leverage semantic features for expert selection. However, our study challenges this notion by demonstrating that positional token information also plays a crucial role in routing decisions. Through extensive empirical analysis, we provide evidence supporting this hypothesis, develop a phenomenological explanation of the observed behavior, and discuss practical implications for MoE-based architectures."},"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":"2504.04444","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-06T11:31:55Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"320afb494f6091e5e2c908762e68061f3cb95fb75db282497f43100d966d6845","abstract_canon_sha256":"0bf072a0f1d4dd22dde0cbed72ba8157294dc4cbf40f229f54fd7e8b5b3cc384"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:45:24.054344Z","signature_b64":"L0C3/EGreqAlX2Yi3y77j/WrTL3Nh1rtYUYWGjHoPObmvfnW0z8MU8jKvHfsppbmOWUcDddsSKYwkNSFiAfHBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d23b913c217a07d9fb223203ffad379b226e5548b3f44ce8e9cb8ed7217ae624","last_reissued_at":"2026-07-05T10:45:24.053873Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:45:24.053873Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On the Spatial Structure of Mixture-of-Experts in Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Daniel Bershatsky, Ivan Oseledets","submitted_at":"2025-04-06T11:31:55Z","abstract_excerpt":"A common assumption is that MoE routers primarily leverage semantic features for expert selection. However, our study challenges this notion by demonstrating that positional token information also plays a crucial role in routing decisions. Through extensive empirical analysis, we provide evidence supporting this hypothesis, develop a phenomenological explanation of the observed behavior, and discuss practical implications for MoE-based architectures."},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.04444","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/2504.04444/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":"2504.04444","created_at":"2026-07-05T10:45:24.053928+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.04444v1","created_at":"2026-07-05T10:45:24.053928+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.04444","created_at":"2026-07-05T10:45:24.053928+00:00"},{"alias_kind":"pith_short_12","alias_value":"2I5ZCPBBPID5","created_at":"2026-07-05T10:45:24.053928+00:00"},{"alias_kind":"pith_short_16","alias_value":"2I5ZCPBBPID5T6ZC","created_at":"2026-07-05T10:45:24.053928+00:00"},{"alias_kind":"pith_short_8","alias_value":"2I5ZCPBB","created_at":"2026-07-05T10:45:24.053928+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.20982","citing_title":"Diagnosing Overhead in Dispatch Operations: Cross-architecture Observatory","ref_index":28,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2I5ZCPBBPID5T6ZCGIB77LJXTM","json":"https://pith.science/pith/2I5ZCPBBPID5T6ZCGIB77LJXTM.json","graph_json":"https://pith.science/api/pith-number/2I5ZCPBBPID5T6ZCGIB77LJXTM/graph.json","events_json":"https://pith.science/api/pith-number/2I5ZCPBBPID5T6ZCGIB77LJXTM/events.json","paper":"https://pith.science/paper/2I5ZCPBB"},"agent_actions":{"view_html":"https://pith.science/pith/2I5ZCPBBPID5T6ZCGIB77LJXTM","download_json":"https://pith.science/pith/2I5ZCPBBPID5T6ZCGIB77LJXTM.json","view_paper":"https://pith.science/paper/2I5ZCPBB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.04444&json=true","fetch_graph":"https://pith.science/api/pith-number/2I5ZCPBBPID5T6ZCGIB77LJXTM/graph.json","fetch_events":"https://pith.science/api/pith-number/2I5ZCPBBPID5T6ZCGIB77LJXTM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2I5ZCPBBPID5T6ZCGIB77LJXTM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2I5ZCPBBPID5T6ZCGIB77LJXTM/action/storage_attestation","attest_author":"https://pith.science/pith/2I5ZCPBBPID5T6ZCGIB77LJXTM/action/author_attestation","sign_citation":"https://pith.science/pith/2I5ZCPBBPID5T6ZCGIB77LJXTM/action/citation_signature","submit_replication":"https://pith.science/pith/2I5ZCPBBPID5T6ZCGIB77LJXTM/action/replication_record"}},"created_at":"2026-07-05T10:45:24.053928+00:00","updated_at":"2026-07-05T10:45:24.053928+00:00"}