{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:IVYN7WDG4FO3K56M4JRKNPI4J4","short_pith_number":"pith:IVYN7WDG","schema_version":"1.0","canonical_sha256":"4570dfd866e15db577cce262a6bd1c4f1d92a456b7e9fd9b44cdadd8118e3762","source":{"kind":"arxiv","id":"2509.02753","version":1},"attestation_state":"computed","paper":{"title":"LExI: Layer-Adaptive Active Experts for Efficient MoE Model Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Krishna Teja Chitty-Venkata, Murali Emani, Sandeep Madireddy, Venkatram Vishwanath","submitted_at":"2025-09-02T19:01:46Z","abstract_excerpt":"Mixture-of-Experts (MoE) models scale efficiently by activating only a subset of experts per token, offering a computationally sparse alternative to dense architectures. While prior post-training optimizations, such as inter- and intra-expert pruning, reduce memory usage they provide limited gains in inference-time compute efficiency. Moreover, existing MoE architectures typically activate a fixed number of experts uniformly across all layers, resulting in redundant computation and suboptimal performance. In this work, we first demonstrate that MoE pruning strategies improve only the memory fo"},"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":"2509.02753","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-02T19:01:46Z","cross_cats_sorted":[],"title_canon_sha256":"69cfd172e9006449ee52a409f2b1263d3a3fe27c004c12b0f2c7662c490933a4","abstract_canon_sha256":"bcc2465eb163612c3bad9755bf2286de31ea1caf52b84b10b59aa6c0427f76c7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:03:57.235206Z","signature_b64":"baNnqlNbtS/3A4uCMLCb9lPzoKP7J+5VOuwyxn6IdcGgKKvh16Lfmyh/zIGvyk3JpOSkNutx9GYhd0flIcL7Dw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4570dfd866e15db577cce262a6bd1c4f1d92a456b7e9fd9b44cdadd8118e3762","last_reissued_at":"2026-07-05T12:03:57.234669Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:03:57.234669Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LExI: Layer-Adaptive Active Experts for Efficient MoE Model Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Krishna Teja Chitty-Venkata, Murali Emani, Sandeep Madireddy, Venkatram Vishwanath","submitted_at":"2025-09-02T19:01:46Z","abstract_excerpt":"Mixture-of-Experts (MoE) models scale efficiently by activating only a subset of experts per token, offering a computationally sparse alternative to dense architectures. While prior post-training optimizations, such as inter- and intra-expert pruning, reduce memory usage they provide limited gains in inference-time compute efficiency. Moreover, existing MoE architectures typically activate a fixed number of experts uniformly across all layers, resulting in redundant computation and suboptimal performance. In this work, we first demonstrate that MoE pruning strategies improve only the memory fo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.02753","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/2509.02753/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":"2509.02753","created_at":"2026-07-05T12:03:57.234727+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.02753v1","created_at":"2026-07-05T12:03:57.234727+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.02753","created_at":"2026-07-05T12:03:57.234727+00:00"},{"alias_kind":"pith_short_12","alias_value":"IVYN7WDG4FO3","created_at":"2026-07-05T12:03:57.234727+00:00"},{"alias_kind":"pith_short_16","alias_value":"IVYN7WDG4FO3K56M","created_at":"2026-07-05T12:03:57.234727+00:00"},{"alias_kind":"pith_short_8","alias_value":"IVYN7WDG","created_at":"2026-07-05T12:03:57.234727+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2510.05497","citing_title":"Patterns behind Chaos: Forecasting Data Movement for Efficient Large-Scale MoE LLM Inference","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2604.08133","citing_title":"Alloc-MoE: Budget-Aware Expert Activation Allocation for Efficient Mixture-of-Experts Inference","ref_index":1,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IVYN7WDG4FO3K56M4JRKNPI4J4","json":"https://pith.science/pith/IVYN7WDG4FO3K56M4JRKNPI4J4.json","graph_json":"https://pith.science/api/pith-number/IVYN7WDG4FO3K56M4JRKNPI4J4/graph.json","events_json":"https://pith.science/api/pith-number/IVYN7WDG4FO3K56M4JRKNPI4J4/events.json","paper":"https://pith.science/paper/IVYN7WDG"},"agent_actions":{"view_html":"https://pith.science/pith/IVYN7WDG4FO3K56M4JRKNPI4J4","download_json":"https://pith.science/pith/IVYN7WDG4FO3K56M4JRKNPI4J4.json","view_paper":"https://pith.science/paper/IVYN7WDG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.02753&json=true","fetch_graph":"https://pith.science/api/pith-number/IVYN7WDG4FO3K56M4JRKNPI4J4/graph.json","fetch_events":"https://pith.science/api/pith-number/IVYN7WDG4FO3K56M4JRKNPI4J4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IVYN7WDG4FO3K56M4JRKNPI4J4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IVYN7WDG4FO3K56M4JRKNPI4J4/action/storage_attestation","attest_author":"https://pith.science/pith/IVYN7WDG4FO3K56M4JRKNPI4J4/action/author_attestation","sign_citation":"https://pith.science/pith/IVYN7WDG4FO3K56M4JRKNPI4J4/action/citation_signature","submit_replication":"https://pith.science/pith/IVYN7WDG4FO3K56M4JRKNPI4J4/action/replication_record"}},"created_at":"2026-07-05T12:03:57.234727+00:00","updated_at":"2026-07-05T12:03:57.234727+00:00"}