{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:KURT36OTTQTXJIBSHGQHPPZYJO","short_pith_number":"pith:KURT36OT","schema_version":"1.0","canonical_sha256":"55233df9d39c2774a03239a077bf384b987253b05ef52d2caa0ad2d49e2025f2","source":{"kind":"arxiv","id":"2006.08217","version":3},"attestation_state":"computed","paper":{"title":"AdamP: Slowing Down the Slowdown for Momentum Optimizers on Scale-invariant Weights","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Byeongho Heo, Dongyoon Han, Gyuwan Kim, Jung-Woo Ha, Sangdoo Yun, Sanghyuk Chun, Seong Joon Oh, Youngjung Uh","submitted_at":"2020-06-15T08:35:15Z","abstract_excerpt":"Normalization techniques are a boon for modern deep learning. They let weights converge more quickly with often better generalization performances. It has been argued that the normalization-induced scale invariance among the weights provides an advantageous ground for gradient descent (GD) optimizers: the effective step sizes are automatically reduced over time, stabilizing the overall training procedure. It is often overlooked, however, that the additional introduction of momentum in GD optimizers results in a far more rapid reduction in effective step sizes for scale-invariant weights, a phe"},"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":"2006.08217","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-06-15T08:35:15Z","cross_cats_sorted":["cs.CV","stat.ML"],"title_canon_sha256":"da0e10e40683ef1e765810f29b960a82c6b6f8fb7554c0783e7d63873e25d879","abstract_canon_sha256":"c97da464bb8756bad709c1363bcc3746213011eaaed366092590d202f5f39f82"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:07:28.707652Z","signature_b64":"rCqkfX4us0sy3LZoepmrO4Wr1pNoaahJt8bu3UKdyHTCcOUMjdAlw5dkuI07Nqro5MhWvyi5FkzAaK9y7oLeCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"55233df9d39c2774a03239a077bf384b987253b05ef52d2caa0ad2d49e2025f2","last_reissued_at":"2026-07-05T02:07:28.707178Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:07:28.707178Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"AdamP: Slowing Down the Slowdown for Momentum Optimizers on Scale-invariant Weights","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Byeongho Heo, Dongyoon Han, Gyuwan Kim, Jung-Woo Ha, Sangdoo Yun, Sanghyuk Chun, Seong Joon Oh, Youngjung Uh","submitted_at":"2020-06-15T08:35:15Z","abstract_excerpt":"Normalization techniques are a boon for modern deep learning. They let weights converge more quickly with often better generalization performances. It has been argued that the normalization-induced scale invariance among the weights provides an advantageous ground for gradient descent (GD) optimizers: the effective step sizes are automatically reduced over time, stabilizing the overall training procedure. It is often overlooked, however, that the additional introduction of momentum in GD optimizers results in a far more rapid reduction in effective step sizes for scale-invariant weights, a phe"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2006.08217","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/2006.08217/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":"2006.08217","created_at":"2026-07-05T02:07:28.707235+00:00"},{"alias_kind":"arxiv_version","alias_value":"2006.08217v3","created_at":"2026-07-05T02:07:28.707235+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2006.08217","created_at":"2026-07-05T02:07:28.707235+00:00"},{"alias_kind":"pith_short_12","alias_value":"KURT36OTTQTX","created_at":"2026-07-05T02:07:28.707235+00:00"},{"alias_kind":"pith_short_16","alias_value":"KURT36OTTQTXJIBS","created_at":"2026-07-05T02:07:28.707235+00:00"},{"alias_kind":"pith_short_8","alias_value":"KURT36OT","created_at":"2026-07-05T02:07:28.707235+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25971","citing_title":"Improving Neural Network Training by Decoupling the Magnitude and Direction of Weight Vectors","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2509.16223","citing_title":"mRadNet: A Compact Radar Object Detector with MetaFormer","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03769","citing_title":"Nora: Normalized Orthogonal Row Alignment for Scalable Matrix Optimizer","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04418","citing_title":"Demystifying Manifold Constraints in LLM Pre-training","ref_index":59,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KURT36OTTQTXJIBSHGQHPPZYJO","json":"https://pith.science/pith/KURT36OTTQTXJIBSHGQHPPZYJO.json","graph_json":"https://pith.science/api/pith-number/KURT36OTTQTXJIBSHGQHPPZYJO/graph.json","events_json":"https://pith.science/api/pith-number/KURT36OTTQTXJIBSHGQHPPZYJO/events.json","paper":"https://pith.science/paper/KURT36OT"},"agent_actions":{"view_html":"https://pith.science/pith/KURT36OTTQTXJIBSHGQHPPZYJO","download_json":"https://pith.science/pith/KURT36OTTQTXJIBSHGQHPPZYJO.json","view_paper":"https://pith.science/paper/KURT36OT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2006.08217&json=true","fetch_graph":"https://pith.science/api/pith-number/KURT36OTTQTXJIBSHGQHPPZYJO/graph.json","fetch_events":"https://pith.science/api/pith-number/KURT36OTTQTXJIBSHGQHPPZYJO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KURT36OTTQTXJIBSHGQHPPZYJO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KURT36OTTQTXJIBSHGQHPPZYJO/action/storage_attestation","attest_author":"https://pith.science/pith/KURT36OTTQTXJIBSHGQHPPZYJO/action/author_attestation","sign_citation":"https://pith.science/pith/KURT36OTTQTXJIBSHGQHPPZYJO/action/citation_signature","submit_replication":"https://pith.science/pith/KURT36OTTQTXJIBSHGQHPPZYJO/action/replication_record"}},"created_at":"2026-07-05T02:07:28.707235+00:00","updated_at":"2026-07-05T02:07:28.707235+00:00"}