{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:QUXHLJ6PILE4SSALYL4RWVZREB","short_pith_number":"pith:QUXHLJ6P","schema_version":"1.0","canonical_sha256":"852e75a7cf42c9c9480bc2f91b57312041d1374e79679a509b4fad4d34d76ef4","source":{"kind":"arxiv","id":"2606.23676","version":1},"attestation_state":"computed","paper":{"title":"Open Problem: Is AdamW Effective Under Heavy-Tailed Noise?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","math.OC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Dingzhi Yu, Hongyi Tao, Lijun Zhang, Luo Luo, Yuanyu Wan","submitted_at":"2026-06-22T17:58:52Z","abstract_excerpt":"AdamW is the de facto optimizer for training large language models (LLMs), yet the theory behind it still lives mostly in finite-variance regimes. This is increasingly unsatisfying, as empirical evidence indicates that stochastic gradient noise in LLM pretraining is typically heavy-tailed. Recent work shows that sign-based optimizers such as Lion and Muon achieve sharp heavy-tailed rates, and that AdaGrad can also converge under heavy-tailed noise. However, no rigorous convergence theory for AdamW has yet been established in this regime. Can AdamW converge under the same heavy-tailed assumptio"},"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":"2606.23676","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-06-22T17:58:52Z","cross_cats_sorted":["cs.AI","math.OC","stat.ML"],"title_canon_sha256":"3fd51354f4542e2b966807afa2791bd5e50976563f0b0c0e3a633db0ab10491e","abstract_canon_sha256":"a225fc760902defa1f7016a51ddae4384553a1294dceae67c4e3baf54668b237"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-23T03:14:34.423931Z","signature_b64":"XBlb8MUhdhaggrMSnB4Hy9/ulCUVGKx5GmE/FeD0427cL1QLky0/tZSZIco/cxVTH+ahY+uqannur7KM8A80Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"852e75a7cf42c9c9480bc2f91b57312041d1374e79679a509b4fad4d34d76ef4","last_reissued_at":"2026-06-23T03:14:34.423572Z","signature_status":"signed_v1","first_computed_at":"2026-06-23T03:14:34.423572Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Open Problem: Is AdamW Effective Under Heavy-Tailed Noise?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","math.OC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Dingzhi Yu, Hongyi Tao, Lijun Zhang, Luo Luo, Yuanyu Wan","submitted_at":"2026-06-22T17:58:52Z","abstract_excerpt":"AdamW is the de facto optimizer for training large language models (LLMs), yet the theory behind it still lives mostly in finite-variance regimes. This is increasingly unsatisfying, as empirical evidence indicates that stochastic gradient noise in LLM pretraining is typically heavy-tailed. Recent work shows that sign-based optimizers such as Lion and Muon achieve sharp heavy-tailed rates, and that AdaGrad can also converge under heavy-tailed noise. However, no rigorous convergence theory for AdamW has yet been established in this regime. Can AdamW converge under the same heavy-tailed assumptio"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.23676","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/2606.23676/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":"2606.23676","created_at":"2026-06-23T03:14:34.423626+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.23676v1","created_at":"2026-06-23T03:14:34.423626+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.23676","created_at":"2026-06-23T03:14:34.423626+00:00"},{"alias_kind":"pith_short_12","alias_value":"QUXHLJ6PILE4","created_at":"2026-06-23T03:14:34.423626+00:00"},{"alias_kind":"pith_short_16","alias_value":"QUXHLJ6PILE4SSAL","created_at":"2026-06-23T03:14:34.423626+00:00"},{"alias_kind":"pith_short_8","alias_value":"QUXHLJ6P","created_at":"2026-06-23T03:14:34.423626+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/QUXHLJ6PILE4SSALYL4RWVZREB","json":"https://pith.science/pith/QUXHLJ6PILE4SSALYL4RWVZREB.json","graph_json":"https://pith.science/api/pith-number/QUXHLJ6PILE4SSALYL4RWVZREB/graph.json","events_json":"https://pith.science/api/pith-number/QUXHLJ6PILE4SSALYL4RWVZREB/events.json","paper":"https://pith.science/paper/QUXHLJ6P"},"agent_actions":{"view_html":"https://pith.science/pith/QUXHLJ6PILE4SSALYL4RWVZREB","download_json":"https://pith.science/pith/QUXHLJ6PILE4SSALYL4RWVZREB.json","view_paper":"https://pith.science/paper/QUXHLJ6P","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.23676&json=true","fetch_graph":"https://pith.science/api/pith-number/QUXHLJ6PILE4SSALYL4RWVZREB/graph.json","fetch_events":"https://pith.science/api/pith-number/QUXHLJ6PILE4SSALYL4RWVZREB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QUXHLJ6PILE4SSALYL4RWVZREB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QUXHLJ6PILE4SSALYL4RWVZREB/action/storage_attestation","attest_author":"https://pith.science/pith/QUXHLJ6PILE4SSALYL4RWVZREB/action/author_attestation","sign_citation":"https://pith.science/pith/QUXHLJ6PILE4SSALYL4RWVZREB/action/citation_signature","submit_replication":"https://pith.science/pith/QUXHLJ6PILE4SSALYL4RWVZREB/action/replication_record"}},"created_at":"2026-06-23T03:14:34.423626+00:00","updated_at":"2026-06-23T03:14:34.423626+00:00"}