{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:4M2DLY3CLNAJMWKBQLNPZ4IPQA","short_pith_number":"pith:4M2DLY3C","schema_version":"1.0","canonical_sha256":"e33435e3625b4096594182dafcf10f802dd9e3f7ff542c8fc7d9f8fabb804236","source":{"kind":"arxiv","id":"2505.05409","version":1},"attestation_state":"computed","paper":{"title":"Hide & Seek: Transformer Symmetries Obscure Sharpness & Riemannian Geometry Finds It","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Felix Dangel, Marvin F. da Silva, Sageev Oore","submitted_at":"2025-05-08T16:51:03Z","abstract_excerpt":"The concept of sharpness has been successfully applied to traditional architectures like MLPs and CNNs to predict their generalization. For transformers, however, recent work reported weak correlation between flatness and generalization. We argue that existing sharpness measures fail for transformers, because they have much richer symmetries in their attention mechanism that induce directions in parameter space along which the network or its loss remain identical. We posit that sharpness must account fully for these symmetries, and thus we redefine it on a quotient manifold that results from q"},"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":"2505.05409","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-08T16:51:03Z","cross_cats_sorted":[],"title_canon_sha256":"d68a29d3f925f52c08e7f673a187dd40483d45ec609dbbeff01f4d8836e7b09f","abstract_canon_sha256":"d6b50be98c75f10c7e676adf8d476b26fb0307945951dc54e3ad3ad5c107404a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:00:23.860660Z","signature_b64":"Q2WhYN5QlMcrLJaaBOeB3hgBslDbi0HA+iOAjbjdu7ekpiWXdtyU3UHQnW3aLrtb98IB9VD1tS1AeX6TeumXCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e33435e3625b4096594182dafcf10f802dd9e3f7ff542c8fc7d9f8fabb804236","last_reissued_at":"2026-07-05T11:00:23.860024Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:00:23.860024Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Hide & Seek: Transformer Symmetries Obscure Sharpness & Riemannian Geometry Finds It","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Felix Dangel, Marvin F. da Silva, Sageev Oore","submitted_at":"2025-05-08T16:51:03Z","abstract_excerpt":"The concept of sharpness has been successfully applied to traditional architectures like MLPs and CNNs to predict their generalization. For transformers, however, recent work reported weak correlation between flatness and generalization. We argue that existing sharpness measures fail for transformers, because they have much richer symmetries in their attention mechanism that induce directions in parameter space along which the network or its loss remain identical. We posit that sharpness must account fully for these symmetries, and thus we redefine it on a quotient manifold that results from q"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.05409","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/2505.05409/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":"2505.05409","created_at":"2026-07-05T11:00:23.860094+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.05409v1","created_at":"2026-07-05T11:00:23.860094+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.05409","created_at":"2026-07-05T11:00:23.860094+00:00"},{"alias_kind":"pith_short_12","alias_value":"4M2DLY3CLNAJ","created_at":"2026-07-05T11:00:23.860094+00:00"},{"alias_kind":"pith_short_16","alias_value":"4M2DLY3CLNAJMWKB","created_at":"2026-07-05T11:00:23.860094+00:00"},{"alias_kind":"pith_short_8","alias_value":"4M2DLY3C","created_at":"2026-07-05T11:00:23.860094+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.18303","citing_title":"A Link between Shock-wave Theory and Symmetry-reduced Stochastic Gradient Descent for Artificial Neural Networks","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27155","citing_title":"Generalizing the Geometry of Model Merging Through Frechet Averages","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27155","citing_title":"Generalizing the Geometry of Model Merging Through Frechet Averages","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4M2DLY3CLNAJMWKBQLNPZ4IPQA","json":"https://pith.science/pith/4M2DLY3CLNAJMWKBQLNPZ4IPQA.json","graph_json":"https://pith.science/api/pith-number/4M2DLY3CLNAJMWKBQLNPZ4IPQA/graph.json","events_json":"https://pith.science/api/pith-number/4M2DLY3CLNAJMWKBQLNPZ4IPQA/events.json","paper":"https://pith.science/paper/4M2DLY3C"},"agent_actions":{"view_html":"https://pith.science/pith/4M2DLY3CLNAJMWKBQLNPZ4IPQA","download_json":"https://pith.science/pith/4M2DLY3CLNAJMWKBQLNPZ4IPQA.json","view_paper":"https://pith.science/paper/4M2DLY3C","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.05409&json=true","fetch_graph":"https://pith.science/api/pith-number/4M2DLY3CLNAJMWKBQLNPZ4IPQA/graph.json","fetch_events":"https://pith.science/api/pith-number/4M2DLY3CLNAJMWKBQLNPZ4IPQA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4M2DLY3CLNAJMWKBQLNPZ4IPQA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4M2DLY3CLNAJMWKBQLNPZ4IPQA/action/storage_attestation","attest_author":"https://pith.science/pith/4M2DLY3CLNAJMWKBQLNPZ4IPQA/action/author_attestation","sign_citation":"https://pith.science/pith/4M2DLY3CLNAJMWKBQLNPZ4IPQA/action/citation_signature","submit_replication":"https://pith.science/pith/4M2DLY3CLNAJMWKBQLNPZ4IPQA/action/replication_record"}},"created_at":"2026-07-05T11:00:23.860094+00:00","updated_at":"2026-07-05T11:00:23.860094+00:00"}