{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:KTBF2AKOPGOD7GATG65VFNYA3G","short_pith_number":"pith:KTBF2AKO","schema_version":"1.0","canonical_sha256":"54c25d014e799c3f981337bb52b700d98a103dc223e24412ce259eec938605da","source":{"kind":"arxiv","id":"2202.00089","version":1},"attestation_state":"computed","paper":{"title":"Understanding AdamW through Proximal Methods and Scale-Freeness","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["math.OC"],"primary_cat":"cs.LG","authors_text":"Ashok Cutkosky, Francesco Orabona, Mingrui Liu, Zhenxun Zhuang","submitted_at":"2022-01-31T21:00:55Z","abstract_excerpt":"Adam has been widely adopted for training deep neural networks due to less hyperparameter tuning and remarkable performance. To improve generalization, Adam is typically used in tandem with a squared $\\ell_2$ regularizer (referred to as Adam-$\\ell_2$). However, even better performance can be obtained with AdamW, which decouples the gradient of the regularizer from the update rule of Adam-$\\ell_2$. Yet, we are still lacking a complete explanation of the advantages of AdamW. In this paper, we tackle this question from both an optimization and an empirical point of view. First, we show how to re-"},"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":"2202.00089","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-01-31T21:00:55Z","cross_cats_sorted":["math.OC"],"title_canon_sha256":"b427ab0616a130d876a71de18d07a540eec46d7ed56d5ae07cab2d9006ea8c8c","abstract_canon_sha256":"9417a84c8f867a32235e37b8deb5273c4a0a678e82c577e1d2aa6b8421d5e30e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:53:18.460049Z","signature_b64":"r1QguVdIAPgjateiJerKgoyyQw5kyFJ0l9FHpTPMZ0VwAx8FpRq2anV7xuw8pDlf7wA4EUL4XDbSOZIRqGeXAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"54c25d014e799c3f981337bb52b700d98a103dc223e24412ce259eec938605da","last_reissued_at":"2026-07-05T03:53:18.459616Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:53:18.459616Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Understanding AdamW through Proximal Methods and Scale-Freeness","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["math.OC"],"primary_cat":"cs.LG","authors_text":"Ashok Cutkosky, Francesco Orabona, Mingrui Liu, Zhenxun Zhuang","submitted_at":"2022-01-31T21:00:55Z","abstract_excerpt":"Adam has been widely adopted for training deep neural networks due to less hyperparameter tuning and remarkable performance. To improve generalization, Adam is typically used in tandem with a squared $\\ell_2$ regularizer (referred to as Adam-$\\ell_2$). However, even better performance can be obtained with AdamW, which decouples the gradient of the regularizer from the update rule of Adam-$\\ell_2$. Yet, we are still lacking a complete explanation of the advantages of AdamW. In this paper, we tackle this question from both an optimization and an empirical point of view. First, we show how to re-"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.00089","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/2202.00089/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":"2202.00089","created_at":"2026-07-05T03:53:18.459669+00:00"},{"alias_kind":"arxiv_version","alias_value":"2202.00089v1","created_at":"2026-07-05T03:53:18.459669+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.00089","created_at":"2026-07-05T03:53:18.459669+00:00"},{"alias_kind":"pith_short_12","alias_value":"KTBF2AKOPGOD","created_at":"2026-07-05T03:53:18.459669+00:00"},{"alias_kind":"pith_short_16","alias_value":"KTBF2AKOPGOD7GAT","created_at":"2026-07-05T03:53:18.459669+00:00"},{"alias_kind":"pith_short_8","alias_value":"KTBF2AKO","created_at":"2026-07-05T03:53:18.459669+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.04418","citing_title":"Demystifying Manifold Constraints in LLM Pre-training","ref_index":57,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KTBF2AKOPGOD7GATG65VFNYA3G","json":"https://pith.science/pith/KTBF2AKOPGOD7GATG65VFNYA3G.json","graph_json":"https://pith.science/api/pith-number/KTBF2AKOPGOD7GATG65VFNYA3G/graph.json","events_json":"https://pith.science/api/pith-number/KTBF2AKOPGOD7GATG65VFNYA3G/events.json","paper":"https://pith.science/paper/KTBF2AKO"},"agent_actions":{"view_html":"https://pith.science/pith/KTBF2AKOPGOD7GATG65VFNYA3G","download_json":"https://pith.science/pith/KTBF2AKOPGOD7GATG65VFNYA3G.json","view_paper":"https://pith.science/paper/KTBF2AKO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2202.00089&json=true","fetch_graph":"https://pith.science/api/pith-number/KTBF2AKOPGOD7GATG65VFNYA3G/graph.json","fetch_events":"https://pith.science/api/pith-number/KTBF2AKOPGOD7GATG65VFNYA3G/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KTBF2AKOPGOD7GATG65VFNYA3G/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KTBF2AKOPGOD7GATG65VFNYA3G/action/storage_attestation","attest_author":"https://pith.science/pith/KTBF2AKOPGOD7GATG65VFNYA3G/action/author_attestation","sign_citation":"https://pith.science/pith/KTBF2AKOPGOD7GATG65VFNYA3G/action/citation_signature","submit_replication":"https://pith.science/pith/KTBF2AKOPGOD7GATG65VFNYA3G/action/replication_record"}},"created_at":"2026-07-05T03:53:18.459669+00:00","updated_at":"2026-07-05T03:53:18.459669+00:00"}