{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:5QE347C6EUITKRJ35EHQ2S7O4P","short_pith_number":"pith:5QE347C6","canonical_record":{"source":{"id":"2605.14521","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T08:05:39Z","cross_cats_sorted":[],"title_canon_sha256":"a0e4549a2e4293bca8e6e7e165bf83178ef392612f9ffb62736c62a0edbf9a8d","abstract_canon_sha256":"53ea40367041f829afd98334caebaf1db0c97d2e18eb075ad08236f7c5a67008"},"schema_version":"1.0"},"canonical_sha256":"ec09be7c5e251135453be90f0d4beee3e58ac61c1287d651d24516862b9b63e3","source":{"kind":"arxiv","id":"2605.14521","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.14521","created_at":"2026-05-17T23:39:06Z"},{"alias_kind":"arxiv_version","alias_value":"2605.14521v1","created_at":"2026-05-17T23:39:06Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.14521","created_at":"2026-05-17T23:39:06Z"},{"alias_kind":"pith_short_12","alias_value":"5QE347C6EUIT","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"5QE347C6EUITKRJ3","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"5QE347C6","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:5QE347C6EUITKRJ35EHQ2S7O4P","target":"record","payload":{"canonical_record":{"source":{"id":"2605.14521","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T08:05:39Z","cross_cats_sorted":[],"title_canon_sha256":"a0e4549a2e4293bca8e6e7e165bf83178ef392612f9ffb62736c62a0edbf9a8d","abstract_canon_sha256":"53ea40367041f829afd98334caebaf1db0c97d2e18eb075ad08236f7c5a67008"},"schema_version":"1.0"},"canonical_sha256":"ec09be7c5e251135453be90f0d4beee3e58ac61c1287d651d24516862b9b63e3","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:06.069090Z","signature_b64":"AwFTzECjocWjJxgHojnf6ZSRbMuqPRex2DI/gBjz23MZ44fUJFZhh49pkizcWREJ87JT6T+K8HFe6LzqADIyBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ec09be7c5e251135453be90f0d4beee3e58ac61c1287d651d24516862b9b63e3","last_reissued_at":"2026-05-17T23:39:06.068337Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:06.068337Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.14521","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:39:06Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"hVCIDjNGRXC0U9pv+WAuED5vjXsIRVdxRmILp5QObIkA7j0dIacdK12ueknfhSy3D2VyD0tNADuWQh2VBHB6Aw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-04T22:45:58.532602Z"},"content_sha256":"1f4f0142bb3bc2fd4ba31dfd539c94909eb67452778d0803c6d7ebb1d59adb94","schema_version":"1.0","event_id":"sha256:1f4f0142bb3bc2fd4ba31dfd539c94909eb67452778d0803c6d7ebb1d59adb94"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:5QE347C6EUITKRJ35EHQ2S7O4P","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Enjoy Your Layer Normalization with the Computational Efficiency of RMSNorm","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Many layer normalizations in standard networks can be folded exactly into upstream layers, allowing precise replacement by faster RMSNorm at inference time with no change in predictions.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Jie Luo, Lei Huang, Wenjun Wu, Yihao Yue, Yizhou Ruan, Yunhao Ni, Yuxin Guo","submitted_at":"2026-05-14T08:05:39Z","abstract_excerpt":"Layer normalization (LN) is a fundamental component in modern deep learning, but its per-sample centering and scaling introduce non-negligible inference overhead. RMSNorm improves efficiency by removing the centering operation, yet this may discard benefits associated with centering. This paper propose a framework to determine whether an LN in an arbitrary DNN can be replaced by RMSNorm without changing the model function. The key idea is to fold LN's centering operation into upstream general linear layers by enforcing zero-mean outputs through the column-centered constraint (CCC) and column-b"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our analysis shows that many LNs in widely used architectures are foldable, enabling exact inference-time conversion and end-to-end acceleration of 2% to 12% without changing model predictions.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the column-centered constraint and column-based weight centering can be enforced on upstream linear layers without changing the overall model function or requiring major retraining adjustments.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A framework to identify and convert foldable layer normalizations to RMSNorm for exact equivalence and faster inference in deep neural networks.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Many layer normalizations in standard networks can be folded exactly into upstream layers, allowing precise replacement by faster RMSNorm at inference time with no change in predictions.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"e7d43e93072e5cf9d8a911dce2d2f798c6956efb580769046b7dd99ddb8bea2e"},"source":{"id":"2605.14521","kind":"arxiv","version":1},"verdict":{"id":"a44b25b9-597b-431b-a9bc-ecb364d786fe","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T02:19:31.026410Z","strongest_claim":"Our analysis shows that many LNs in widely used architectures are foldable, enabling exact inference-time conversion and end-to-end acceleration of 2% to 12% without changing model predictions.","one_line_summary":"A framework to identify and convert foldable layer normalizations to RMSNorm for exact equivalence and faster inference in deep neural networks.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the column-centered constraint and column-based weight centering can be enforced on upstream linear layers without changing the overall model function or requiring major retraining adjustments.","pith_extraction_headline":"Many layer normalizations in standard networks can be folded exactly into upstream layers, allowing precise replacement by faster RMSNorm at inference time with no change in predictions."},"references":{"count":105,"sample":[{"doi":"","year":null,"title":"ICML , year =","work_id":"df41788f-0406-4dc5-ac74-ada95b438b18","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1906,"title":"Contrastive multiview coding","work_id":"e401469c-d91d-466c-94b1-63fd434d83fd","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Chun-Hsiao Yeh, Yubei Chen , howpublished=","work_id":"edf8cb6c-af95-41e6-9f31-f66699f5f5b4","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year=","work_id":"c4fdb5db-88ac-40bb-bb65-60e6d770a1e3","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2016,"title":"Layer Normalization , author=. 2016 , eprint=","work_id":"2d03abe2-0f55-41a8-9cb9-e8d3f4376824","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":105,"snapshot_sha256":"3ba48788aafa0d212856b4bf1655894d92d172b21f0cb9a604736b0b9b5b64bb","internal_anchors":9},"formal_canon":{"evidence_count":1,"snapshot_sha256":"dd548e576b15087da03c7345b5a9f49ff45fa8ce93ad41a3cd41c092b4bd1065"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"a44b25b9-597b-431b-a9bc-ecb364d786fe"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:39:06Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"4QYLMk+6RQ0t9yfJVdJdRMYrKWY8FD4BrDXISOYo+g3WbZTMTDaYz/1BS1uP47gwKCxJOwe/3ibRxFoDn+x6DQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-04T22:45:58.533111Z"},"content_sha256":"7eea4c5d54bf5721ffb455b1ac0a0879da626a3a793a6ece7b4ddf96e1010479","schema_version":"1.0","event_id":"sha256:7eea4c5d54bf5721ffb455b1ac0a0879da626a3a793a6ece7b4ddf96e1010479"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/5QE347C6EUITKRJ35EHQ2S7O4P/bundle.json","state_url":"https://pith.science/pith/5QE347C6EUITKRJ35EHQ2S7O4P/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/5QE347C6EUITKRJ35EHQ2S7O4P/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-04T22:45:58Z","links":{"resolver":"https://pith.science/pith/5QE347C6EUITKRJ35EHQ2S7O4P","bundle":"https://pith.science/pith/5QE347C6EUITKRJ35EHQ2S7O4P/bundle.json","state":"https://pith.science/pith/5QE347C6EUITKRJ35EHQ2S7O4P/state.json","well_known_bundle":"https://pith.science/.well-known/pith/5QE347C6EUITKRJ35EHQ2S7O4P/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:5QE347C6EUITKRJ35EHQ2S7O4P","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"53ea40367041f829afd98334caebaf1db0c97d2e18eb075ad08236f7c5a67008","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T08:05:39Z","title_canon_sha256":"a0e4549a2e4293bca8e6e7e165bf83178ef392612f9ffb62736c62a0edbf9a8d"},"schema_version":"1.0","source":{"id":"2605.14521","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.14521","created_at":"2026-05-17T23:39:06Z"},{"alias_kind":"arxiv_version","alias_value":"2605.14521v1","created_at":"2026-05-17T23:39:06Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.14521","created_at":"2026-05-17T23:39:06Z"},{"alias_kind":"pith_short_12","alias_value":"5QE347C6EUIT","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"5QE347C6EUITKRJ3","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"5QE347C6","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:7eea4c5d54bf5721ffb455b1ac0a0879da626a3a793a6ece7b4ddf96e1010479","target":"graph","created_at":"2026-05-17T23:39:06Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Our analysis shows that many LNs in widely used architectures are foldable, enabling exact inference-time conversion and end-to-end acceleration of 2% to 12% without changing model predictions."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the column-centered constraint and column-based weight centering can be enforced on upstream linear layers without changing the overall model function or requiring major retraining adjustments."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"A framework to identify and convert foldable layer normalizations to RMSNorm for exact equivalence and faster inference in deep neural networks."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Many layer normalizations in standard networks can be folded exactly into upstream layers, allowing precise replacement by faster RMSNorm at inference time with no change in predictions."}],"snapshot_sha256":"e7d43e93072e5cf9d8a911dce2d2f798c6956efb580769046b7dd99ddb8bea2e"},"formal_canon":{"evidence_count":1,"snapshot_sha256":"dd548e576b15087da03c7345b5a9f49ff45fa8ce93ad41a3cd41c092b4bd1065"},"paper":{"abstract_excerpt":"Layer normalization (LN) is a fundamental component in modern deep learning, but its per-sample centering and scaling introduce non-negligible inference overhead. RMSNorm improves efficiency by removing the centering operation, yet this may discard benefits associated with centering. This paper propose a framework to determine whether an LN in an arbitrary DNN can be replaced by RMSNorm without changing the model function. The key idea is to fold LN's centering operation into upstream general linear layers by enforcing zero-mean outputs through the column-centered constraint (CCC) and column-b","authors_text":"Jie Luo, Lei Huang, Wenjun Wu, Yihao Yue, Yizhou Ruan, Yunhao Ni, Yuxin Guo","cross_cats":[],"headline":"Many layer normalizations in standard networks can be folded exactly into upstream layers, allowing precise replacement by faster RMSNorm at inference time with no change in predictions.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T08:05:39Z","title":"Enjoy Your Layer Normalization with the Computational Efficiency of RMSNorm"},"references":{"count":105,"internal_anchors":9,"resolved_work":105,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"ICML , year =","work_id":"df41788f-0406-4dc5-ac74-ada95b438b18","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Contrastive multiview coding","work_id":"e401469c-d91d-466c-94b1-63fd434d83fd","year":1906},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Chun-Hsiao Yeh, Yubei Chen , howpublished=","work_id":"edf8cb6c-af95-41e6-9f31-f66699f5f5b4","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year=","work_id":"c4fdb5db-88ac-40bb-bb65-60e6d770a1e3","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Layer Normalization , author=. 2016 , eprint=","work_id":"2d03abe2-0f55-41a8-9cb9-e8d3f4376824","year":2016}],"snapshot_sha256":"3ba48788aafa0d212856b4bf1655894d92d172b21f0cb9a604736b0b9b5b64bb"},"source":{"id":"2605.14521","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-15T02:19:31.026410Z","id":"a44b25b9-597b-431b-a9bc-ecb364d786fe","model_set":{"reader":"grok-4.3"},"one_line_summary":"A framework to identify and convert foldable layer normalizations to RMSNorm for exact equivalence and faster inference in deep neural networks.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Many layer normalizations in standard networks can be folded exactly into upstream layers, allowing precise replacement by faster RMSNorm at inference time with no change in predictions.","strongest_claim":"Our analysis shows that many LNs in widely used architectures are foldable, enabling exact inference-time conversion and end-to-end acceleration of 2% to 12% without changing model predictions.","weakest_assumption":"That the column-centered constraint and column-based weight centering can be enforced on upstream linear layers without changing the overall model function or requiring major retraining adjustments."}},"verdict_id":"a44b25b9-597b-431b-a9bc-ecb364d786fe"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:1f4f0142bb3bc2fd4ba31dfd539c94909eb67452778d0803c6d7ebb1d59adb94","target":"record","created_at":"2026-05-17T23:39:06Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"53ea40367041f829afd98334caebaf1db0c97d2e18eb075ad08236f7c5a67008","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T08:05:39Z","title_canon_sha256":"a0e4549a2e4293bca8e6e7e165bf83178ef392612f9ffb62736c62a0edbf9a8d"},"schema_version":"1.0","source":{"id":"2605.14521","kind":"arxiv","version":1}},"canonical_sha256":"ec09be7c5e251135453be90f0d4beee3e58ac61c1287d651d24516862b9b63e3","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"ec09be7c5e251135453be90f0d4beee3e58ac61c1287d651d24516862b9b63e3","first_computed_at":"2026-05-17T23:39:06.068337Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:39:06.068337Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"AwFTzECjocWjJxgHojnf6ZSRbMuqPRex2DI/gBjz23MZ44fUJFZhh49pkizcWREJ87JT6T+K8HFe6LzqADIyBQ==","signature_status":"signed_v1","signed_at":"2026-05-17T23:39:06.069090Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.14521","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:1f4f0142bb3bc2fd4ba31dfd539c94909eb67452778d0803c6d7ebb1d59adb94","sha256:7eea4c5d54bf5721ffb455b1ac0a0879da626a3a793a6ece7b4ddf96e1010479"],"state_sha256":"9f68773f505c08590cba2ecbec50168cba64c18b6a999eaef42f8801033f3190"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"w09D7kOw1hiEEjGaEKQ03QA9g9bRog3CBfQeWwIg9MnFIjXxxuF3Xd9r0hNGx5r7f3gE/J6G74+4b3/um7zBDw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-04T22:45:58.535519Z","bundle_sha256":"ee2f19a5548aa83e53a9515058f140e77cc6bd0d3ef810b6851cd3ce16bd1228"}}