{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:D3GL7KQK27KFRAZUEPDRFUC57B","short_pith_number":"pith:D3GL7KQK","schema_version":"1.0","canonical_sha256":"1eccbfaa0ad7d458833423c712d05df85d2b2f4933a780cc2ae8edeab7e8111d","source":{"kind":"arxiv","id":"2501.19057","version":1},"attestation_state":"computed","paper":{"title":"TeZO: Empowering the Low-Rankness on the Temporal Dimension in the Zeroth-Order Optimization for Fine-tuning LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Dacheng Tao, Liang Ding, Li Shen, Tiansheng Huang, Yan Sun","submitted_at":"2025-01-31T11:34:03Z","abstract_excerpt":"Zeroth-order optimization (ZO) has demonstrated remarkable promise in efficient fine-tuning tasks for Large Language Models (LLMs). In particular, recent advances incorporate the low-rankness of gradients, introducing low-rank ZO estimators to further reduce GPU memory consumption. However, most existing works focus solely on the low-rankness of each individual gradient, overlooking a broader property shared by all gradients throughout the training, i.e., all gradients approximately reside within a similar subspace. In this paper, we consider two factors together and propose a novel low-rank Z"},"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":"2501.19057","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-31T11:34:03Z","cross_cats_sorted":[],"title_canon_sha256":"f8b7d0db70c59316310b30e50029df5d226187420fd264de6ea187cfd40df666","abstract_canon_sha256":"914a92b9721e609ab7d8b5a517665ab4229142e6e1aa291fe50bc0804d2a7d2c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:07:55.622798Z","signature_b64":"FGq626CInwO0R2FU7L1wfKvNAlsYYT+/ut7JBaWq2TDszuW6Ler317z7pVgttpBThjX+GNHqczZ3mUEUq05GDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1eccbfaa0ad7d458833423c712d05df85d2b2f4933a780cc2ae8edeab7e8111d","last_reissued_at":"2026-07-05T10:07:55.622340Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:07:55.622340Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TeZO: Empowering the Low-Rankness on the Temporal Dimension in the Zeroth-Order Optimization for Fine-tuning LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Dacheng Tao, Liang Ding, Li Shen, Tiansheng Huang, Yan Sun","submitted_at":"2025-01-31T11:34:03Z","abstract_excerpt":"Zeroth-order optimization (ZO) has demonstrated remarkable promise in efficient fine-tuning tasks for Large Language Models (LLMs). In particular, recent advances incorporate the low-rankness of gradients, introducing low-rank ZO estimators to further reduce GPU memory consumption. However, most existing works focus solely on the low-rankness of each individual gradient, overlooking a broader property shared by all gradients throughout the training, i.e., all gradients approximately reside within a similar subspace. In this paper, we consider two factors together and propose a novel low-rank Z"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.19057","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/2501.19057/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":"2501.19057","created_at":"2026-07-05T10:07:55.622399+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.19057v1","created_at":"2026-07-05T10:07:55.622399+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.19057","created_at":"2026-07-05T10:07:55.622399+00:00"},{"alias_kind":"pith_short_12","alias_value":"D3GL7KQK27KF","created_at":"2026-07-05T10:07:55.622399+00:00"},{"alias_kind":"pith_short_16","alias_value":"D3GL7KQK27KFRAZU","created_at":"2026-07-05T10:07:55.622399+00:00"},{"alias_kind":"pith_short_8","alias_value":"D3GL7KQK","created_at":"2026-07-05T10:07:55.622399+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.00650","citing_title":"AdaMeZO: Adam-style Zeroth-Order Optimizer for LLM Fine-tuning Without Maintaining the Moments","ref_index":59,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12968","citing_title":"Evolution of Optimization Methods: Algorithms, Scenarios, and Evaluations","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18264","citing_title":"Universally Empowering Zeroth-Order Optimization via Adaptive Layer-wise Sampling","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/D3GL7KQK27KFRAZUEPDRFUC57B","json":"https://pith.science/pith/D3GL7KQK27KFRAZUEPDRFUC57B.json","graph_json":"https://pith.science/api/pith-number/D3GL7KQK27KFRAZUEPDRFUC57B/graph.json","events_json":"https://pith.science/api/pith-number/D3GL7KQK27KFRAZUEPDRFUC57B/events.json","paper":"https://pith.science/paper/D3GL7KQK"},"agent_actions":{"view_html":"https://pith.science/pith/D3GL7KQK27KFRAZUEPDRFUC57B","download_json":"https://pith.science/pith/D3GL7KQK27KFRAZUEPDRFUC57B.json","view_paper":"https://pith.science/paper/D3GL7KQK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.19057&json=true","fetch_graph":"https://pith.science/api/pith-number/D3GL7KQK27KFRAZUEPDRFUC57B/graph.json","fetch_events":"https://pith.science/api/pith-number/D3GL7KQK27KFRAZUEPDRFUC57B/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/D3GL7KQK27KFRAZUEPDRFUC57B/action/timestamp_anchor","attest_storage":"https://pith.science/pith/D3GL7KQK27KFRAZUEPDRFUC57B/action/storage_attestation","attest_author":"https://pith.science/pith/D3GL7KQK27KFRAZUEPDRFUC57B/action/author_attestation","sign_citation":"https://pith.science/pith/D3GL7KQK27KFRAZUEPDRFUC57B/action/citation_signature","submit_replication":"https://pith.science/pith/D3GL7KQK27KFRAZUEPDRFUC57B/action/replication_record"}},"created_at":"2026-07-05T10:07:55.622399+00:00","updated_at":"2026-07-05T10:07:55.622399+00:00"}