{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:MZZX7DTG7ARN4QELSHY7PFN2N4","short_pith_number":"pith:MZZX7DTG","schema_version":"1.0","canonical_sha256":"66737f8e66f822de408b91f1f795ba6f024d84f4605780e278a8abc1ae37ed60","source":{"kind":"arxiv","id":"2406.07337","version":1},"attestation_state":"computed","paper":{"title":"Transferring Knowledge from Large Foundation Models to Small Downstream Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Andrew Gordon Wilson, Boran Han, Danielle C. Maddix, Shikai Qiu, Shuai Zhang, Yuyang Wang","submitted_at":"2024-06-11T15:06:15Z","abstract_excerpt":"How do we transfer the relevant knowledge from ever larger foundation models into small, task-specific downstream models that can run at much lower costs? Standard transfer learning using pre-trained weights as the initialization transfers limited information and commits us to often massive pre-trained architectures. This procedure also precludes combining multiple pre-trained models that learn complementary information. To address these shortcomings, we introduce Adaptive Feature Transfer (AFT). Instead of transferring weights, AFT operates purely on features, thereby decoupling the choice of"},"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":"2406.07337","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-06-11T15:06:15Z","cross_cats_sorted":[],"title_canon_sha256":"2ba626d47ec2a66ed84cb88103002644ceafb4df2f9f9df1e5f3ef48b49e2a62","abstract_canon_sha256":"91eef85d399b7e76b8a7474eedeb1a9ebda184d014019e511f3d645a1abba985"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:30:21.806379Z","signature_b64":"WZ4igvPB4+49Q/zLZ12XrGg3dil6xDQnMO4VJGsFHv3PH60ztsRWqdd7MAqjvUKho+AvuSgyx+IDCmsR+0o9Ag==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"66737f8e66f822de408b91f1f795ba6f024d84f4605780e278a8abc1ae37ed60","last_reissued_at":"2026-07-05T08:30:21.805903Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:30:21.805903Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Transferring Knowledge from Large Foundation Models to Small Downstream Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Andrew Gordon Wilson, Boran Han, Danielle C. Maddix, Shikai Qiu, Shuai Zhang, Yuyang Wang","submitted_at":"2024-06-11T15:06:15Z","abstract_excerpt":"How do we transfer the relevant knowledge from ever larger foundation models into small, task-specific downstream models that can run at much lower costs? Standard transfer learning using pre-trained weights as the initialization transfers limited information and commits us to often massive pre-trained architectures. This procedure also precludes combining multiple pre-trained models that learn complementary information. To address these shortcomings, we introduce Adaptive Feature Transfer (AFT). Instead of transferring weights, AFT operates purely on features, thereby decoupling the choice of"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.07337","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/2406.07337/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":"2406.07337","created_at":"2026-07-05T08:30:21.805962+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.07337v1","created_at":"2026-07-05T08:30:21.805962+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.07337","created_at":"2026-07-05T08:30:21.805962+00:00"},{"alias_kind":"pith_short_12","alias_value":"MZZX7DTG7ARN","created_at":"2026-07-05T08:30:21.805962+00:00"},{"alias_kind":"pith_short_16","alias_value":"MZZX7DTG7ARN4QEL","created_at":"2026-07-05T08:30:21.805962+00:00"},{"alias_kind":"pith_short_8","alias_value":"MZZX7DTG","created_at":"2026-07-05T08:30:21.805962+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2501.09009","citing_title":"Towards Fast, Specialized Machine Learning Force Fields: Distilling Foundation Models via Energy Hessians","ref_index":18,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MZZX7DTG7ARN4QELSHY7PFN2N4","json":"https://pith.science/pith/MZZX7DTG7ARN4QELSHY7PFN2N4.json","graph_json":"https://pith.science/api/pith-number/MZZX7DTG7ARN4QELSHY7PFN2N4/graph.json","events_json":"https://pith.science/api/pith-number/MZZX7DTG7ARN4QELSHY7PFN2N4/events.json","paper":"https://pith.science/paper/MZZX7DTG"},"agent_actions":{"view_html":"https://pith.science/pith/MZZX7DTG7ARN4QELSHY7PFN2N4","download_json":"https://pith.science/pith/MZZX7DTG7ARN4QELSHY7PFN2N4.json","view_paper":"https://pith.science/paper/MZZX7DTG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.07337&json=true","fetch_graph":"https://pith.science/api/pith-number/MZZX7DTG7ARN4QELSHY7PFN2N4/graph.json","fetch_events":"https://pith.science/api/pith-number/MZZX7DTG7ARN4QELSHY7PFN2N4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MZZX7DTG7ARN4QELSHY7PFN2N4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MZZX7DTG7ARN4QELSHY7PFN2N4/action/storage_attestation","attest_author":"https://pith.science/pith/MZZX7DTG7ARN4QELSHY7PFN2N4/action/author_attestation","sign_citation":"https://pith.science/pith/MZZX7DTG7ARN4QELSHY7PFN2N4/action/citation_signature","submit_replication":"https://pith.science/pith/MZZX7DTG7ARN4QELSHY7PFN2N4/action/replication_record"}},"created_at":"2026-07-05T08:30:21.805962+00:00","updated_at":"2026-07-05T08:30:21.805962+00:00"}