{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:CAXIZ7G3SP5UM2MBUZIPNCW5QR","short_pith_number":"pith:CAXIZ7G3","schema_version":"1.0","canonical_sha256":"102e8cfcdb93fb466981a650f68add84651ae6ee95ecbb4619a38a3582fab2d0","source":{"kind":"arxiv","id":"2606.04212","version":1},"attestation_state":"computed","paper":{"title":"Edge of Stability Selectively Shapes Learning Across the Data Distribution","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Anakha Ganesh, Pierfrancesco Beneventano, Shauna Kwag, Tomaso Poggio","submitted_at":"2026-06-02T20:58:40Z","abstract_excerpt":"Existing analyses of the edge of stability (EoS) treat it as a global property of optimization. We show that it is also selective: the stability constraint redistributes learning across subsets of the training distribution, amplifying progress on some groups while suppressing progress on others. Using a branching intervention that enters or exits the EoS regime from the same training state, we causally demonstrate this trade-off and identify two necessary conditions for a group to benefit. First, its aggregate gradient must align with the top Hessian eigenvector. We isolate this mechanism with"},"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":"2606.04212","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-02T20:58:40Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"99a4540a3482c22473c18b2793c7c176bfbb6b6c1318694db2cc07d07a092dd5","abstract_canon_sha256":"5e8db6f884208d4ce7cfb21e40822212f157352b606f52a6c7ac8ee38beefb2f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-04T01:08:58.818214Z","signature_b64":"EQdAkwN2WOFxuOm7VSQKFFlhxwq60KDOiaJW3HaAQHj5dt+jE56p9bbXbeDdrpVcQJH+AMz7NSVcs/JJNGRJCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"102e8cfcdb93fb466981a650f68add84651ae6ee95ecbb4619a38a3582fab2d0","last_reissued_at":"2026-06-04T01:08:58.817734Z","signature_status":"signed_v1","first_computed_at":"2026-06-04T01:08:58.817734Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Edge of Stability Selectively Shapes Learning Across the Data Distribution","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Anakha Ganesh, Pierfrancesco Beneventano, Shauna Kwag, Tomaso Poggio","submitted_at":"2026-06-02T20:58:40Z","abstract_excerpt":"Existing analyses of the edge of stability (EoS) treat it as a global property of optimization. We show that it is also selective: the stability constraint redistributes learning across subsets of the training distribution, amplifying progress on some groups while suppressing progress on others. Using a branching intervention that enters or exits the EoS regime from the same training state, we causally demonstrate this trade-off and identify two necessary conditions for a group to benefit. First, its aggregate gradient must align with the top Hessian eigenvector. We isolate this mechanism with"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.04212","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/2606.04212/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":"2606.04212","created_at":"2026-06-04T01:08:58.817807+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.04212v1","created_at":"2026-06-04T01:08:58.817807+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.04212","created_at":"2026-06-04T01:08:58.817807+00:00"},{"alias_kind":"pith_short_12","alias_value":"CAXIZ7G3SP5U","created_at":"2026-06-04T01:08:58.817807+00:00"},{"alias_kind":"pith_short_16","alias_value":"CAXIZ7G3SP5UM2MB","created_at":"2026-06-04T01:08:58.817807+00:00"},{"alias_kind":"pith_short_8","alias_value":"CAXIZ7G3","created_at":"2026-06-04T01:08:58.817807+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/CAXIZ7G3SP5UM2MBUZIPNCW5QR","json":"https://pith.science/pith/CAXIZ7G3SP5UM2MBUZIPNCW5QR.json","graph_json":"https://pith.science/api/pith-number/CAXIZ7G3SP5UM2MBUZIPNCW5QR/graph.json","events_json":"https://pith.science/api/pith-number/CAXIZ7G3SP5UM2MBUZIPNCW5QR/events.json","paper":"https://pith.science/paper/CAXIZ7G3"},"agent_actions":{"view_html":"https://pith.science/pith/CAXIZ7G3SP5UM2MBUZIPNCW5QR","download_json":"https://pith.science/pith/CAXIZ7G3SP5UM2MBUZIPNCW5QR.json","view_paper":"https://pith.science/paper/CAXIZ7G3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.04212&json=true","fetch_graph":"https://pith.science/api/pith-number/CAXIZ7G3SP5UM2MBUZIPNCW5QR/graph.json","fetch_events":"https://pith.science/api/pith-number/CAXIZ7G3SP5UM2MBUZIPNCW5QR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CAXIZ7G3SP5UM2MBUZIPNCW5QR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CAXIZ7G3SP5UM2MBUZIPNCW5QR/action/storage_attestation","attest_author":"https://pith.science/pith/CAXIZ7G3SP5UM2MBUZIPNCW5QR/action/author_attestation","sign_citation":"https://pith.science/pith/CAXIZ7G3SP5UM2MBUZIPNCW5QR/action/citation_signature","submit_replication":"https://pith.science/pith/CAXIZ7G3SP5UM2MBUZIPNCW5QR/action/replication_record"}},"created_at":"2026-06-04T01:08:58.817807+00:00","updated_at":"2026-06-04T01:08:58.817807+00:00"}