{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:ZG7HFXVBOEYCIUBC2KNR4LQBRO","short_pith_number":"pith:ZG7HFXVB","schema_version":"1.0","canonical_sha256":"c9be72dea17130245022d29b1e2e018bbf87a23e4b53e23e356410404e61500c","source":{"kind":"arxiv","id":"2304.09856","version":1},"attestation_state":"computed","paper":{"title":"LipsFormer: Introducing Lipschitz Continuity to Vision Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Jianan Wang, Lei Zhang, Xianbiao Qi, Yihao Chen, Yukai Shi","submitted_at":"2023-04-19T17:59:39Z","abstract_excerpt":"We present a Lipschitz continuous Transformer, called LipsFormer, to pursue training stability both theoretically and empirically for Transformer-based models. In contrast to previous practical tricks that address training instability by learning rate warmup, layer normalization, attention formulation, and weight initialization, we show that Lipschitz continuity is a more essential property to ensure training stability. In LipsFormer, we replace unstable Transformer component modules with Lipschitz continuous counterparts: CenterNorm instead of LayerNorm, spectral initialization instead of Xav"},"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":"2304.09856","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2023-04-19T17:59:39Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"c8310d28c46445daccdcb4a3ee24677d8308857f125fdbde02a5762a23980ad7","abstract_canon_sha256":"0d06885fa197b16fef5ad9255cad41adcb12a163da21d887e8627fd2eb82f1a0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:02:43.006832Z","signature_b64":"LhMhabVKey3WvrTjCt6nnOzCUQBGo+I0XmP8umAJoT3P8YMXqGRJWvLGL9gWabSZAnYDwWj/OpMgmdU/cSslCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c9be72dea17130245022d29b1e2e018bbf87a23e4b53e23e356410404e61500c","last_reissued_at":"2026-07-05T06:02:43.006480Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:02:43.006480Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LipsFormer: Introducing Lipschitz Continuity to Vision Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Jianan Wang, Lei Zhang, Xianbiao Qi, Yihao Chen, Yukai Shi","submitted_at":"2023-04-19T17:59:39Z","abstract_excerpt":"We present a Lipschitz continuous Transformer, called LipsFormer, to pursue training stability both theoretically and empirically for Transformer-based models. In contrast to previous practical tricks that address training instability by learning rate warmup, layer normalization, attention formulation, and weight initialization, we show that Lipschitz continuity is a more essential property to ensure training stability. In LipsFormer, we replace unstable Transformer component modules with Lipschitz continuous counterparts: CenterNorm instead of LayerNorm, spectral initialization instead of Xav"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2304.09856","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/2304.09856/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":"2304.09856","created_at":"2026-07-05T06:02:43.006538+00:00"},{"alias_kind":"arxiv_version","alias_value":"2304.09856v1","created_at":"2026-07-05T06:02:43.006538+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2304.09856","created_at":"2026-07-05T06:02:43.006538+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZG7HFXVBOEYC","created_at":"2026-07-05T06:02:43.006538+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZG7HFXVBOEYCIUBC","created_at":"2026-07-05T06:02:43.006538+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZG7HFXVB","created_at":"2026-07-05T06:02:43.006538+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.07925","citing_title":"Sinkhorn doubly stochastic attention rank decay analysis","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15838","citing_title":"Reversible Residual Normalization Alleviates Spatio-Temporal Distribution Shift","ref_index":9,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZG7HFXVBOEYCIUBC2KNR4LQBRO","json":"https://pith.science/pith/ZG7HFXVBOEYCIUBC2KNR4LQBRO.json","graph_json":"https://pith.science/api/pith-number/ZG7HFXVBOEYCIUBC2KNR4LQBRO/graph.json","events_json":"https://pith.science/api/pith-number/ZG7HFXVBOEYCIUBC2KNR4LQBRO/events.json","paper":"https://pith.science/paper/ZG7HFXVB"},"agent_actions":{"view_html":"https://pith.science/pith/ZG7HFXVBOEYCIUBC2KNR4LQBRO","download_json":"https://pith.science/pith/ZG7HFXVBOEYCIUBC2KNR4LQBRO.json","view_paper":"https://pith.science/paper/ZG7HFXVB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2304.09856&json=true","fetch_graph":"https://pith.science/api/pith-number/ZG7HFXVBOEYCIUBC2KNR4LQBRO/graph.json","fetch_events":"https://pith.science/api/pith-number/ZG7HFXVBOEYCIUBC2KNR4LQBRO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZG7HFXVBOEYCIUBC2KNR4LQBRO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZG7HFXVBOEYCIUBC2KNR4LQBRO/action/storage_attestation","attest_author":"https://pith.science/pith/ZG7HFXVBOEYCIUBC2KNR4LQBRO/action/author_attestation","sign_citation":"https://pith.science/pith/ZG7HFXVBOEYCIUBC2KNR4LQBRO/action/citation_signature","submit_replication":"https://pith.science/pith/ZG7HFXVBOEYCIUBC2KNR4LQBRO/action/replication_record"}},"created_at":"2026-07-05T06:02:43.006538+00:00","updated_at":"2026-07-05T06:02:43.006538+00:00"}