{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:OMLDDTTWUGZS2OPXHIDTAPM4WB","short_pith_number":"pith:OMLDDTTW","schema_version":"1.0","canonical_sha256":"731631ce76a1b32d39f73a07303d9cb05e948738c4b663330b51478048aff590","source":{"kind":"arxiv","id":"2411.09702","version":1},"attestation_state":"computed","paper":{"title":"On the Surprising Effectiveness of Attention Transfer for Vision Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.NE"],"primary_cat":"cs.LG","authors_text":"Alexander C. Li, Beidi Chen, Deepak Pathak, Xinlei Chen, Yuandong Tian","submitted_at":"2024-11-14T18:59:40Z","abstract_excerpt":"Conventional wisdom suggests that pre-training Vision Transformers (ViT) improves downstream performance by learning useful representations. Is this actually true? We investigate this question and find that the features and representations learned during pre-training are not essential. Surprisingly, using only the attention patterns from pre-training (i.e., guiding how information flows between tokens) is sufficient for models to learn high quality features from scratch and achieve comparable downstream performance. We show this by introducing a simple method called attention transfer, where o"},"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":"2411.09702","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-11-14T18:59:40Z","cross_cats_sorted":["cs.AI","cs.CV","cs.NE"],"title_canon_sha256":"3f7c033518846350a6e68e88a4686b69f7ef278839d936f04d0632b7eebb0725","abstract_canon_sha256":"d1ca1a5769530f3d86c30b4584aa344a59f48a447263a4525d4d8e03275f4128"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:35:30.990455Z","signature_b64":"pK87kGhPvioWoUerwppsHYSZCoZgmciK+AK4kI8biVx3BuQDGEl7cokMIGMvP7i+hl161Gjnqxq0Aza5PK0hBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"731631ce76a1b32d39f73a07303d9cb05e948738c4b663330b51478048aff590","last_reissued_at":"2026-07-05T09:35:30.989967Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:35:30.989967Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On the Surprising Effectiveness of Attention Transfer for Vision Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.NE"],"primary_cat":"cs.LG","authors_text":"Alexander C. Li, Beidi Chen, Deepak Pathak, Xinlei Chen, Yuandong Tian","submitted_at":"2024-11-14T18:59:40Z","abstract_excerpt":"Conventional wisdom suggests that pre-training Vision Transformers (ViT) improves downstream performance by learning useful representations. Is this actually true? We investigate this question and find that the features and representations learned during pre-training are not essential. Surprisingly, using only the attention patterns from pre-training (i.e., guiding how information flows between tokens) is sufficient for models to learn high quality features from scratch and achieve comparable downstream performance. We show this by introducing a simple method called attention transfer, where o"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.09702","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/2411.09702/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":"2411.09702","created_at":"2026-07-05T09:35:30.990032+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.09702v1","created_at":"2026-07-05T09:35:30.990032+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.09702","created_at":"2026-07-05T09:35:30.990032+00:00"},{"alias_kind":"pith_short_12","alias_value":"OMLDDTTWUGZS","created_at":"2026-07-05T09:35:30.990032+00:00"},{"alias_kind":"pith_short_16","alias_value":"OMLDDTTWUGZS2OPX","created_at":"2026-07-05T09:35:30.990032+00:00"},{"alias_kind":"pith_short_8","alias_value":"OMLDDTTW","created_at":"2026-07-05T09:35:30.990032+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/OMLDDTTWUGZS2OPXHIDTAPM4WB","json":"https://pith.science/pith/OMLDDTTWUGZS2OPXHIDTAPM4WB.json","graph_json":"https://pith.science/api/pith-number/OMLDDTTWUGZS2OPXHIDTAPM4WB/graph.json","events_json":"https://pith.science/api/pith-number/OMLDDTTWUGZS2OPXHIDTAPM4WB/events.json","paper":"https://pith.science/paper/OMLDDTTW"},"agent_actions":{"view_html":"https://pith.science/pith/OMLDDTTWUGZS2OPXHIDTAPM4WB","download_json":"https://pith.science/pith/OMLDDTTWUGZS2OPXHIDTAPM4WB.json","view_paper":"https://pith.science/paper/OMLDDTTW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.09702&json=true","fetch_graph":"https://pith.science/api/pith-number/OMLDDTTWUGZS2OPXHIDTAPM4WB/graph.json","fetch_events":"https://pith.science/api/pith-number/OMLDDTTWUGZS2OPXHIDTAPM4WB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OMLDDTTWUGZS2OPXHIDTAPM4WB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OMLDDTTWUGZS2OPXHIDTAPM4WB/action/storage_attestation","attest_author":"https://pith.science/pith/OMLDDTTWUGZS2OPXHIDTAPM4WB/action/author_attestation","sign_citation":"https://pith.science/pith/OMLDDTTWUGZS2OPXHIDTAPM4WB/action/citation_signature","submit_replication":"https://pith.science/pith/OMLDDTTWUGZS2OPXHIDTAPM4WB/action/replication_record"}},"created_at":"2026-07-05T09:35:30.990032+00:00","updated_at":"2026-07-05T09:35:30.990032+00:00"}