{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:6JDLNKJQBSDVHMASZP3JINFQY6","short_pith_number":"pith:6JDLNKJQ","schema_version":"1.0","canonical_sha256":"f246b6a9300c8753b012cbf69434b0c798397ebcb345b5b39b7a1ef1c311f5b9","source":{"kind":"arxiv","id":"2412.10663","version":2},"attestation_state":"computed","paper":{"title":"Memory-Efficient 4-bit Preconditioned Stochastic Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","math.OC"],"primary_cat":"cs.LG","authors_text":"Jingyang Li, Kim-Chuan Toh, Kuangyu Ding, Pan Zhou","submitted_at":"2024-12-14T03:32:54Z","abstract_excerpt":"Preconditioned stochastic optimization algorithms, exemplified by Shampoo, outperform first-order optimizers by offering theoretical convergence benefits and practical gains in large-scale neural network training. However, they incur substantial memory overhead due to the storage demands of non-diagonal preconditioning matrices. To address this, we introduce 4-bit quantization for Shampoo's preconditioners. We introduce two key methods: First, we apply Cholesky decomposition followed by quantization of the Cholesky factors, reducing memory usage by leveraging their lower triangular structure w"},"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":"2412.10663","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-12-14T03:32:54Z","cross_cats_sorted":["cs.CV","math.OC"],"title_canon_sha256":"12c6cf4d6e6a1024c924312f4ffe34a16120ac8a53ae81fe62136825a766a00e","abstract_canon_sha256":"fb8408995c961b3e564af880adee8909f209a4266e735bcc1d441368f64732f2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:29:28.518807Z","signature_b64":"/Ag+o2MC/tKJjRXusgZh0gVvhT6pkIVCElDrRDz3jHvgjUchFNvGzOfOtqmxiHsCeeqUy7Tuj/eY0ns+KlX0Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f246b6a9300c8753b012cbf69434b0c798397ebcb345b5b39b7a1ef1c311f5b9","last_reissued_at":"2026-07-05T10:29:28.518110Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:29:28.518110Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Memory-Efficient 4-bit Preconditioned Stochastic Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","math.OC"],"primary_cat":"cs.LG","authors_text":"Jingyang Li, Kim-Chuan Toh, Kuangyu Ding, Pan Zhou","submitted_at":"2024-12-14T03:32:54Z","abstract_excerpt":"Preconditioned stochastic optimization algorithms, exemplified by Shampoo, outperform first-order optimizers by offering theoretical convergence benefits and practical gains in large-scale neural network training. However, they incur substantial memory overhead due to the storage demands of non-diagonal preconditioning matrices. To address this, we introduce 4-bit quantization for Shampoo's preconditioners. We introduce two key methods: First, we apply Cholesky decomposition followed by quantization of the Cholesky factors, reducing memory usage by leveraging their lower triangular structure w"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.10663","kind":"arxiv","version":2},"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/2412.10663/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":"2412.10663","created_at":"2026-07-05T10:29:28.518198+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.10663v2","created_at":"2026-07-05T10:29:28.518198+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.10663","created_at":"2026-07-05T10:29:28.518198+00:00"},{"alias_kind":"pith_short_12","alias_value":"6JDLNKJQBSDV","created_at":"2026-07-05T10:29:28.518198+00:00"},{"alias_kind":"pith_short_16","alias_value":"6JDLNKJQBSDVHMAS","created_at":"2026-07-05T10:29:28.518198+00:00"},{"alias_kind":"pith_short_8","alias_value":"6JDLNKJQ","created_at":"2026-07-05T10:29:28.518198+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/6JDLNKJQBSDVHMASZP3JINFQY6","json":"https://pith.science/pith/6JDLNKJQBSDVHMASZP3JINFQY6.json","graph_json":"https://pith.science/api/pith-number/6JDLNKJQBSDVHMASZP3JINFQY6/graph.json","events_json":"https://pith.science/api/pith-number/6JDLNKJQBSDVHMASZP3JINFQY6/events.json","paper":"https://pith.science/paper/6JDLNKJQ"},"agent_actions":{"view_html":"https://pith.science/pith/6JDLNKJQBSDVHMASZP3JINFQY6","download_json":"https://pith.science/pith/6JDLNKJQBSDVHMASZP3JINFQY6.json","view_paper":"https://pith.science/paper/6JDLNKJQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.10663&json=true","fetch_graph":"https://pith.science/api/pith-number/6JDLNKJQBSDVHMASZP3JINFQY6/graph.json","fetch_events":"https://pith.science/api/pith-number/6JDLNKJQBSDVHMASZP3JINFQY6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6JDLNKJQBSDVHMASZP3JINFQY6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6JDLNKJQBSDVHMASZP3JINFQY6/action/storage_attestation","attest_author":"https://pith.science/pith/6JDLNKJQBSDVHMASZP3JINFQY6/action/author_attestation","sign_citation":"https://pith.science/pith/6JDLNKJQBSDVHMASZP3JINFQY6/action/citation_signature","submit_replication":"https://pith.science/pith/6JDLNKJQBSDVHMASZP3JINFQY6/action/replication_record"}},"created_at":"2026-07-05T10:29:28.518198+00:00","updated_at":"2026-07-05T10:29:28.518198+00:00"}