{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KDOTSNB6MOTXWOROHIIMV4AN5M","short_pith_number":"pith:KDOTSNB6","schema_version":"1.0","canonical_sha256":"50dd39343e63a77b3a2e3a10caf00deb27151f97d06dba68a990d6f103175965","source":{"kind":"arxiv","id":"2605.26647","version":1},"attestation_state":"computed","paper":{"title":"More Expressive Feedforward Layers: Part I. Token-Adaptive Mixing of Activations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Jinbo Wang, Kai Shen, Mingze Wang, Shu Zhong, Yikuan Xia","submitted_at":"2026-05-26T07:30:53Z","abstract_excerpt":"Feedforward network (FFN) layers account for a large fraction of parameters and nonlinear expressivity in Transformer-based large language models (LLMs). Despite the evolution from ReLU and GELU to gated variants such as SwiGLU, most FFN designs still use a single fixed activation function, applying the same nonlinear transformation to all tokens. In this work, we propose Mixture of Activations (MoA), a token-adaptive FFN design that mixes a dictionary of activation functions using lightweight input-dependent gates while sharing the same linear projections. As an input-independent counterpart,"},"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":"2605.26647","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-26T07:30:53Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"ce2e0a86d94bbd60606c3e43d7cc6bf5c20a54a675b7a49bc37bf8cbc2e44a34","abstract_canon_sha256":"5118be827b9482fabf3a1ece50b4809253ae0c1cdf8d26b22cef39e1a96df143"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T01:06:03.986516Z","signature_b64":"mWIInFVnPpuFIBuIvCRi906obW2w8XHX9nMcq1y1wCgLw3Awb11LjpfAe2TQeropTDTycnfAPtn91WobjhEICg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"50dd39343e63a77b3a2e3a10caf00deb27151f97d06dba68a990d6f103175965","last_reissued_at":"2026-05-27T01:06:03.985639Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T01:06:03.985639Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"More Expressive Feedforward Layers: Part I. Token-Adaptive Mixing of Activations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Jinbo Wang, Kai Shen, Mingze Wang, Shu Zhong, Yikuan Xia","submitted_at":"2026-05-26T07:30:53Z","abstract_excerpt":"Feedforward network (FFN) layers account for a large fraction of parameters and nonlinear expressivity in Transformer-based large language models (LLMs). Despite the evolution from ReLU and GELU to gated variants such as SwiGLU, most FFN designs still use a single fixed activation function, applying the same nonlinear transformation to all tokens. In this work, we propose Mixture of Activations (MoA), a token-adaptive FFN design that mixes a dictionary of activation functions using lightweight input-dependent gates while sharing the same linear projections. As an input-independent counterpart,"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.26647","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/2605.26647/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":"2605.26647","created_at":"2026-05-27T01:06:03.985781+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.26647v1","created_at":"2026-05-27T01:06:03.985781+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.26647","created_at":"2026-05-27T01:06:03.985781+00:00"},{"alias_kind":"pith_short_12","alias_value":"KDOTSNB6MOTX","created_at":"2026-05-27T01:06:03.985781+00:00"},{"alias_kind":"pith_short_16","alias_value":"KDOTSNB6MOTXWORO","created_at":"2026-05-27T01:06:03.985781+00:00"},{"alias_kind":"pith_short_8","alias_value":"KDOTSNB6","created_at":"2026-05-27T01:06:03.985781+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/KDOTSNB6MOTXWOROHIIMV4AN5M","json":"https://pith.science/pith/KDOTSNB6MOTXWOROHIIMV4AN5M.json","graph_json":"https://pith.science/api/pith-number/KDOTSNB6MOTXWOROHIIMV4AN5M/graph.json","events_json":"https://pith.science/api/pith-number/KDOTSNB6MOTXWOROHIIMV4AN5M/events.json","paper":"https://pith.science/paper/KDOTSNB6"},"agent_actions":{"view_html":"https://pith.science/pith/KDOTSNB6MOTXWOROHIIMV4AN5M","download_json":"https://pith.science/pith/KDOTSNB6MOTXWOROHIIMV4AN5M.json","view_paper":"https://pith.science/paper/KDOTSNB6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.26647&json=true","fetch_graph":"https://pith.science/api/pith-number/KDOTSNB6MOTXWOROHIIMV4AN5M/graph.json","fetch_events":"https://pith.science/api/pith-number/KDOTSNB6MOTXWOROHIIMV4AN5M/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KDOTSNB6MOTXWOROHIIMV4AN5M/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KDOTSNB6MOTXWOROHIIMV4AN5M/action/storage_attestation","attest_author":"https://pith.science/pith/KDOTSNB6MOTXWOROHIIMV4AN5M/action/author_attestation","sign_citation":"https://pith.science/pith/KDOTSNB6MOTXWOROHIIMV4AN5M/action/citation_signature","submit_replication":"https://pith.science/pith/KDOTSNB6MOTXWOROHIIMV4AN5M/action/replication_record"}},"created_at":"2026-05-27T01:06:03.985781+00:00","updated_at":"2026-05-27T01:06:03.985781+00:00"}