{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:BDRAQH7HQHMYCGY7K54CE7FURP","short_pith_number":"pith:BDRAQH7H","schema_version":"1.0","canonical_sha256":"08e2081fe781d9811b1f5778227cb48bd70ed1b768bdea0e9b5012ef431fca9b","source":{"kind":"arxiv","id":"2401.15207","version":3},"attestation_state":"computed","paper":{"title":"HiFT: A Hierarchical Full Parameter Fine-Tuning Strategy","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Daling Wang, Hinrich Sch\\\"utze, Qian Li, Shi Feng, Tong Liu, Yifei Zhang, Yiqun Zhang, Yongkang Liu","submitted_at":"2024-01-26T21:14:32Z","abstract_excerpt":"Full-parameter fine-tuning has become the go-to choice for adapting language models (LMs) to downstream tasks due to its excellent performance. As LMs grow in size, fine-tuning the full parameters of LMs requires a prohibitively large amount of GPU memory. Existing approaches utilize zeroth-order optimizer to conserve GPU memory, which can potentially compromise the performance of LMs as non-zero order optimizers tend to converge more readily on most downstream tasks. In this paper, we propose a novel optimizer-independent end-to-end hierarchical fine-tuning strategy, HiFT, which only updates "},"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":"2401.15207","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-01-26T21:14:32Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"3a71ba492060f869ad363e3666ce5347a238225680e41cac65ca0f4e3744149b","abstract_canon_sha256":"0af553dda978b704ca1c894a2c0b7b42f7a20d04d32a55d43862ab5b6603228f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:32:35.353799Z","signature_b64":"w2kxUJi7cBK8x/nwXunpbkwernViKarYRM1RDU0HJ+/VSDSUw/A6EN2rw4XNr8U/6xF4oigia2tZ17auj70GCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"08e2081fe781d9811b1f5778227cb48bd70ed1b768bdea0e9b5012ef431fca9b","last_reissued_at":"2026-07-05T08:32:35.353341Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:32:35.353341Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"HiFT: A Hierarchical Full Parameter Fine-Tuning Strategy","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Daling Wang, Hinrich Sch\\\"utze, Qian Li, Shi Feng, Tong Liu, Yifei Zhang, Yiqun Zhang, Yongkang Liu","submitted_at":"2024-01-26T21:14:32Z","abstract_excerpt":"Full-parameter fine-tuning has become the go-to choice for adapting language models (LMs) to downstream tasks due to its excellent performance. As LMs grow in size, fine-tuning the full parameters of LMs requires a prohibitively large amount of GPU memory. Existing approaches utilize zeroth-order optimizer to conserve GPU memory, which can potentially compromise the performance of LMs as non-zero order optimizers tend to converge more readily on most downstream tasks. In this paper, we propose a novel optimizer-independent end-to-end hierarchical fine-tuning strategy, HiFT, which only updates "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.15207","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/2401.15207/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":"2401.15207","created_at":"2026-07-05T08:32:35.353399+00:00"},{"alias_kind":"arxiv_version","alias_value":"2401.15207v3","created_at":"2026-07-05T08:32:35.353399+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.15207","created_at":"2026-07-05T08:32:35.353399+00:00"},{"alias_kind":"pith_short_12","alias_value":"BDRAQH7HQHMY","created_at":"2026-07-05T08:32:35.353399+00:00"},{"alias_kind":"pith_short_16","alias_value":"BDRAQH7HQHMYCGY7","created_at":"2026-07-05T08:32:35.353399+00:00"},{"alias_kind":"pith_short_8","alias_value":"BDRAQH7H","created_at":"2026-07-05T08:32:35.353399+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.22633","citing_title":"H2Tune: Federated Foundation Model Fine-Tuning with Hybrid Heterogeneity","ref_index":29,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BDRAQH7HQHMYCGY7K54CE7FURP","json":"https://pith.science/pith/BDRAQH7HQHMYCGY7K54CE7FURP.json","graph_json":"https://pith.science/api/pith-number/BDRAQH7HQHMYCGY7K54CE7FURP/graph.json","events_json":"https://pith.science/api/pith-number/BDRAQH7HQHMYCGY7K54CE7FURP/events.json","paper":"https://pith.science/paper/BDRAQH7H"},"agent_actions":{"view_html":"https://pith.science/pith/BDRAQH7HQHMYCGY7K54CE7FURP","download_json":"https://pith.science/pith/BDRAQH7HQHMYCGY7K54CE7FURP.json","view_paper":"https://pith.science/paper/BDRAQH7H","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2401.15207&json=true","fetch_graph":"https://pith.science/api/pith-number/BDRAQH7HQHMYCGY7K54CE7FURP/graph.json","fetch_events":"https://pith.science/api/pith-number/BDRAQH7HQHMYCGY7K54CE7FURP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BDRAQH7HQHMYCGY7K54CE7FURP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BDRAQH7HQHMYCGY7K54CE7FURP/action/storage_attestation","attest_author":"https://pith.science/pith/BDRAQH7HQHMYCGY7K54CE7FURP/action/author_attestation","sign_citation":"https://pith.science/pith/BDRAQH7HQHMYCGY7K54CE7FURP/action/citation_signature","submit_replication":"https://pith.science/pith/BDRAQH7HQHMYCGY7K54CE7FURP/action/replication_record"}},"created_at":"2026-07-05T08:32:35.353399+00:00","updated_at":"2026-07-05T08:32:35.353399+00:00"}