{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:4AGKOU6HEROVDXWI3EQOJYHJ2I","short_pith_number":"pith:4AGKOU6H","schema_version":"1.0","canonical_sha256":"e00ca753c7245d51dec8d920e4e0e9d23424ddf80c4bb3bbc21228c6c0b754b1","source":{"kind":"arxiv","id":"2208.01313","version":1},"attestation_state":"computed","paper":{"title":"Unified Normalization for Accelerating and Stabilizing Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.MM"],"primary_cat":"cs.CV","authors_text":"Chaoxiang Lan, Jun Xiao, Kai Zhang, Qiming Yang, Shiliang Pu, Wenming Tan, Zheyang Li, Zhi Yang","submitted_at":"2022-08-02T08:41:31Z","abstract_excerpt":"Solid results from Transformers have made them prevailing architectures in various natural language and vision tasks. As a default component in Transformers, Layer Normalization (LN) normalizes activations within each token to boost the robustness. However, LN requires on-the-fly statistics calculation in inference as well as division and square root operations, leading to inefficiency on hardware. What is more, replacing LN with other hardware-efficient normalization schemes (e.g., Batch Normalization) results in inferior performance, even collapse in training. We find that this dilemma is ca"},"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":"2208.01313","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2022-08-02T08:41:31Z","cross_cats_sorted":["cs.MM"],"title_canon_sha256":"a3e90eaf71f64e795290074083c80fed2045032ce27eb763480737ed1cba8976","abstract_canon_sha256":"eae9d7822fc5b8c8b6a2854991fd12d75889b17d65203af36b1f15567d6aa57c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:45:32.183723Z","signature_b64":"RICXFqHBiJ/LTnC9DK6kLIbZ9qD+dhhQ2Sm4VKMTEEhD+czd/tXYGlqf8J+hqDwe0s5uGSwESMag0IPHIx7mCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e00ca753c7245d51dec8d920e4e0e9d23424ddf80c4bb3bbc21228c6c0b754b1","last_reissued_at":"2026-07-05T04:45:32.183249Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:45:32.183249Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Unified Normalization for Accelerating and Stabilizing Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.MM"],"primary_cat":"cs.CV","authors_text":"Chaoxiang Lan, Jun Xiao, Kai Zhang, Qiming Yang, Shiliang Pu, Wenming Tan, Zheyang Li, Zhi Yang","submitted_at":"2022-08-02T08:41:31Z","abstract_excerpt":"Solid results from Transformers have made them prevailing architectures in various natural language and vision tasks. As a default component in Transformers, Layer Normalization (LN) normalizes activations within each token to boost the robustness. However, LN requires on-the-fly statistics calculation in inference as well as division and square root operations, leading to inefficiency on hardware. What is more, replacing LN with other hardware-efficient normalization schemes (e.g., Batch Normalization) results in inferior performance, even collapse in training. We find that this dilemma is ca"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2208.01313","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/2208.01313/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":"2208.01313","created_at":"2026-07-05T04:45:32.183303+00:00"},{"alias_kind":"arxiv_version","alias_value":"2208.01313v1","created_at":"2026-07-05T04:45:32.183303+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2208.01313","created_at":"2026-07-05T04:45:32.183303+00:00"},{"alias_kind":"pith_short_12","alias_value":"4AGKOU6HEROV","created_at":"2026-07-05T04:45:32.183303+00:00"},{"alias_kind":"pith_short_16","alias_value":"4AGKOU6HEROVDXWI","created_at":"2026-07-05T04:45:32.183303+00:00"},{"alias_kind":"pith_short_8","alias_value":"4AGKOU6H","created_at":"2026-07-05T04:45:32.183303+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2306.14048","citing_title":"H$_2$O: Heavy-Hitter Oracle for Efficient Generative Inference of Large Language Models","ref_index":91,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4AGKOU6HEROVDXWI3EQOJYHJ2I","json":"https://pith.science/pith/4AGKOU6HEROVDXWI3EQOJYHJ2I.json","graph_json":"https://pith.science/api/pith-number/4AGKOU6HEROVDXWI3EQOJYHJ2I/graph.json","events_json":"https://pith.science/api/pith-number/4AGKOU6HEROVDXWI3EQOJYHJ2I/events.json","paper":"https://pith.science/paper/4AGKOU6H"},"agent_actions":{"view_html":"https://pith.science/pith/4AGKOU6HEROVDXWI3EQOJYHJ2I","download_json":"https://pith.science/pith/4AGKOU6HEROVDXWI3EQOJYHJ2I.json","view_paper":"https://pith.science/paper/4AGKOU6H","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2208.01313&json=true","fetch_graph":"https://pith.science/api/pith-number/4AGKOU6HEROVDXWI3EQOJYHJ2I/graph.json","fetch_events":"https://pith.science/api/pith-number/4AGKOU6HEROVDXWI3EQOJYHJ2I/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4AGKOU6HEROVDXWI3EQOJYHJ2I/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4AGKOU6HEROVDXWI3EQOJYHJ2I/action/storage_attestation","attest_author":"https://pith.science/pith/4AGKOU6HEROVDXWI3EQOJYHJ2I/action/author_attestation","sign_citation":"https://pith.science/pith/4AGKOU6HEROVDXWI3EQOJYHJ2I/action/citation_signature","submit_replication":"https://pith.science/pith/4AGKOU6HEROVDXWI3EQOJYHJ2I/action/replication_record"}},"created_at":"2026-07-05T04:45:32.183303+00:00","updated_at":"2026-07-05T04:45:32.183303+00:00"}