{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:YW23VQT665DLL7ZLISFS6AXWQJ","short_pith_number":"pith:YW23VQT6","schema_version":"1.0","canonical_sha256":"c5b5bac27ef746b5ff2b448b2f02f6824102b46b987b773a822ee8c80bd65fba","source":{"kind":"arxiv","id":"2606.00761","version":1},"attestation_state":"computed","paper":{"title":"Confidence-Adaptive SwiGLU for Mixture-of-Experts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Liangli Zhen, Rick Siow Mong Goh, Shaohua Li, Xiaobing Sun, Xiuchao Sui, Yong Liu, Yuhang Wu","submitted_at":"2026-05-30T14:58:52Z","abstract_excerpt":"SwiGLU has become a standard gated activation in modern Transformer MLPs, yet its gate sharpness -- the smoothness and selectivity of the gating function -- is typically fixed throughout training. In this work, we propose Confidence-Aware SwiGLU ($\\kappa$-SwiGLU), a variant of SwiGLU for Mixture-of-Experts (MoE) models that adjusts expert gate sharpness according to token-level routing confidence. Specifically, $\\kappa$-SwiGLU parameterizes the SiLU gate sharpness coefficient as a learnable function of the router logit, enabling each expert gate unit to interpolate between smooth, broadly acti"},"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":"2606.00761","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-30T14:58:52Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"248778a827c56431924e891224cb4f7c296ede1244889e236e056b1828e0b0b3","abstract_canon_sha256":"6f65849a815df0dbc2f8416109dc1dc16ee726dcee426e450418446feb9cc78b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T01:04:04.990993Z","signature_b64":"meBjMrt05PrmJrulv5GYaSZlzbFf2wcaq3Bn1yndZ8dB60RqhYOGpgoGHtfp2ZO0KbGphDXoJL9LR3PiuIBzCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c5b5bac27ef746b5ff2b448b2f02f6824102b46b987b773a822ee8c80bd65fba","last_reissued_at":"2026-06-02T01:04:04.990623Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T01:04:04.990623Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Confidence-Adaptive SwiGLU for Mixture-of-Experts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Liangli Zhen, Rick Siow Mong Goh, Shaohua Li, Xiaobing Sun, Xiuchao Sui, Yong Liu, Yuhang Wu","submitted_at":"2026-05-30T14:58:52Z","abstract_excerpt":"SwiGLU has become a standard gated activation in modern Transformer MLPs, yet its gate sharpness -- the smoothness and selectivity of the gating function -- is typically fixed throughout training. In this work, we propose Confidence-Aware SwiGLU ($\\kappa$-SwiGLU), a variant of SwiGLU for Mixture-of-Experts (MoE) models that adjusts expert gate sharpness according to token-level routing confidence. Specifically, $\\kappa$-SwiGLU parameterizes the SiLU gate sharpness coefficient as a learnable function of the router logit, enabling each expert gate unit to interpolate between smooth, broadly acti"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.00761","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/2606.00761/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":"2606.00761","created_at":"2026-06-02T01:04:04.990681+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.00761v1","created_at":"2026-06-02T01:04:04.990681+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.00761","created_at":"2026-06-02T01:04:04.990681+00:00"},{"alias_kind":"pith_short_12","alias_value":"YW23VQT665DL","created_at":"2026-06-02T01:04:04.990681+00:00"},{"alias_kind":"pith_short_16","alias_value":"YW23VQT665DLL7ZL","created_at":"2026-06-02T01:04:04.990681+00:00"},{"alias_kind":"pith_short_8","alias_value":"YW23VQT6","created_at":"2026-06-02T01:04:04.990681+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/YW23VQT665DLL7ZLISFS6AXWQJ","json":"https://pith.science/pith/YW23VQT665DLL7ZLISFS6AXWQJ.json","graph_json":"https://pith.science/api/pith-number/YW23VQT665DLL7ZLISFS6AXWQJ/graph.json","events_json":"https://pith.science/api/pith-number/YW23VQT665DLL7ZLISFS6AXWQJ/events.json","paper":"https://pith.science/paper/YW23VQT6"},"agent_actions":{"view_html":"https://pith.science/pith/YW23VQT665DLL7ZLISFS6AXWQJ","download_json":"https://pith.science/pith/YW23VQT665DLL7ZLISFS6AXWQJ.json","view_paper":"https://pith.science/paper/YW23VQT6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.00761&json=true","fetch_graph":"https://pith.science/api/pith-number/YW23VQT665DLL7ZLISFS6AXWQJ/graph.json","fetch_events":"https://pith.science/api/pith-number/YW23VQT665DLL7ZLISFS6AXWQJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YW23VQT665DLL7ZLISFS6AXWQJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YW23VQT665DLL7ZLISFS6AXWQJ/action/storage_attestation","attest_author":"https://pith.science/pith/YW23VQT665DLL7ZLISFS6AXWQJ/action/author_attestation","sign_citation":"https://pith.science/pith/YW23VQT665DLL7ZLISFS6AXWQJ/action/citation_signature","submit_replication":"https://pith.science/pith/YW23VQT665DLL7ZLISFS6AXWQJ/action/replication_record"}},"created_at":"2026-06-02T01:04:04.990681+00:00","updated_at":"2026-06-02T01:04:04.990681+00:00"}