{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:CTADMPSUWHKYMQ3ZPAGTGPGMMV","short_pith_number":"pith:CTADMPSU","schema_version":"1.0","canonical_sha256":"14c0363e54b1d5864379780d333ccc6553b1968a306211af177db9981d6dfe3e","source":{"kind":"arxiv","id":"2410.13849","version":3},"attestation_state":"computed","paper":{"title":"From Gradient Clipping to Normalization for Heavy Tailed SGD","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","stat.ML"],"primary_cat":"math.OC","authors_text":"Florian H\\\"ubler, Ilyas Fatkhullin, Niao He","submitted_at":"2024-10-17T17:59:01Z","abstract_excerpt":"Recent empirical evidence indicates that many machine learning applications involve heavy-tailed gradient noise, which challenges the standard assumptions of bounded variance in stochastic optimization. Gradient clipping has emerged as a popular tool to handle this heavy-tailed noise, as it achieves good performance in this setting both theoretically and practically. However, our current theoretical understanding of non-convex gradient clipping has three main shortcomings. First, the theory hinges on large, increasing clipping thresholds, which are in stark contrast to the small constant clipp"},"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":"2410.13849","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"math.OC","submitted_at":"2024-10-17T17:59:01Z","cross_cats_sorted":["cs.LG","stat.ML"],"title_canon_sha256":"c9f89eedac83b4f31f9921cf092f8cceb57b1c0699e5f7c58860d4d6ebb06248","abstract_canon_sha256":"9915a3a2caf3a444dd149a57793e13d2585cd1aec3bb11ab631a412b997433c2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:34:00.673420Z","signature_b64":"ogcx6xzaYZ3xz+eZd0wrDipGEp0d1qsQKvyUJEiiNBLEOPuKTH1w1HNdGQA25jkZXgzBdYmXSUu8T7nSSsLZCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"14c0363e54b1d5864379780d333ccc6553b1968a306211af177db9981d6dfe3e","last_reissued_at":"2026-07-05T11:34:00.672793Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:34:00.672793Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"From Gradient Clipping to Normalization for Heavy Tailed SGD","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","stat.ML"],"primary_cat":"math.OC","authors_text":"Florian H\\\"ubler, Ilyas Fatkhullin, Niao He","submitted_at":"2024-10-17T17:59:01Z","abstract_excerpt":"Recent empirical evidence indicates that many machine learning applications involve heavy-tailed gradient noise, which challenges the standard assumptions of bounded variance in stochastic optimization. Gradient clipping has emerged as a popular tool to handle this heavy-tailed noise, as it achieves good performance in this setting both theoretically and practically. However, our current theoretical understanding of non-convex gradient clipping has three main shortcomings. First, the theory hinges on large, increasing clipping thresholds, which are in stark contrast to the small constant clipp"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.13849","kind":"arxiv","version":3},"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/2410.13849/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":"2410.13849","created_at":"2026-07-05T11:34:00.672866+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.13849v3","created_at":"2026-07-05T11:34:00.672866+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.13849","created_at":"2026-07-05T11:34:00.672866+00:00"},{"alias_kind":"pith_short_12","alias_value":"CTADMPSUWHKY","created_at":"2026-07-05T11:34:00.672866+00:00"},{"alias_kind":"pith_short_16","alias_value":"CTADMPSUWHKYMQ3Z","created_at":"2026-07-05T11:34:00.672866+00:00"},{"alias_kind":"pith_short_8","alias_value":"CTADMPSU","created_at":"2026-07-05T11:34:00.672866+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.19811","citing_title":"LionMuon: Alternating Spectral and Sign Descent for Efficient Training","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.24513","citing_title":"Zeroth-Order Nonconvex Nonsmooth Optimization with Heavy-Tailed Noise","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27733","citing_title":"Can Entry-Wise Clipping Give Spectral Control of Stochastic Gradients?","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2505.03736","citing_title":"Decentralized Nonconvex Optimization under Heavy-Tailed Noise: Normalization and Optimal Convergence","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19811","citing_title":"LionMuon: Alternating Spectral and Sign Descent for Efficient Training","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17394","citing_title":"Stochastic Zeroth-Order Optimization Under Heavy-Tailed Noise","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11838","citing_title":"Gradient Clipping Beyond Vector Norms: A Spectral Approach for Matrix-Valued Parameters","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06884","citing_title":"Muon with Nesterov Momentum: Heavy-Tailed Noise and (Randomized) Inexact Polar Decomposition","ref_index":19,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CTADMPSUWHKYMQ3ZPAGTGPGMMV","json":"https://pith.science/pith/CTADMPSUWHKYMQ3ZPAGTGPGMMV.json","graph_json":"https://pith.science/api/pith-number/CTADMPSUWHKYMQ3ZPAGTGPGMMV/graph.json","events_json":"https://pith.science/api/pith-number/CTADMPSUWHKYMQ3ZPAGTGPGMMV/events.json","paper":"https://pith.science/paper/CTADMPSU"},"agent_actions":{"view_html":"https://pith.science/pith/CTADMPSUWHKYMQ3ZPAGTGPGMMV","download_json":"https://pith.science/pith/CTADMPSUWHKYMQ3ZPAGTGPGMMV.json","view_paper":"https://pith.science/paper/CTADMPSU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.13849&json=true","fetch_graph":"https://pith.science/api/pith-number/CTADMPSUWHKYMQ3ZPAGTGPGMMV/graph.json","fetch_events":"https://pith.science/api/pith-number/CTADMPSUWHKYMQ3ZPAGTGPGMMV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CTADMPSUWHKYMQ3ZPAGTGPGMMV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CTADMPSUWHKYMQ3ZPAGTGPGMMV/action/storage_attestation","attest_author":"https://pith.science/pith/CTADMPSUWHKYMQ3ZPAGTGPGMMV/action/author_attestation","sign_citation":"https://pith.science/pith/CTADMPSUWHKYMQ3ZPAGTGPGMMV/action/citation_signature","submit_replication":"https://pith.science/pith/CTADMPSUWHKYMQ3ZPAGTGPGMMV/action/replication_record"}},"created_at":"2026-07-05T11:34:00.672866+00:00","updated_at":"2026-07-05T11:34:00.672866+00:00"}