{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:UY5EMRF45ZZMCPBCUQSAJKTOOQ","short_pith_number":"pith:UY5EMRF4","schema_version":"1.0","canonical_sha256":"a63a4644bcee72c13c22a42404aa6e740aa390bf7e9c0b78d920e7f0ed556ecf","source":{"kind":"arxiv","id":"2401.14391","version":2},"attestation_state":"computed","paper":{"title":"Rethinking Patch Dependence for Masked Autoencoders","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Adam Yala, Alexei A. Efros, Baifeng Shi, Ken Goldberg, Letian Fu, Long Lian, Renhao Wang, Trevor Darrell, Xudong Wang","submitted_at":"2024-01-25T18:49:57Z","abstract_excerpt":"In this work, we examine the impact of inter-patch dependencies in the decoder of masked autoencoders (MAE) on representation learning. We decompose the decoding mechanism for masked reconstruction into self-attention between mask tokens and cross-attention between masked and visible tokens. Our findings reveal that MAE reconstructs coherent images from visible patches not through interactions between patches in the decoder but by learning a global representation within the encoder. This discovery leads us to propose a simple visual pretraining framework: cross-attention masked autoencoders (C"},"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":"2401.14391","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-01-25T18:49:57Z","cross_cats_sorted":[],"title_canon_sha256":"93e34e1fc5657002e895aa1609ec398b84ecf117ded0b5271bd46686149d8433","abstract_canon_sha256":"d9ada401340f568b5ae8b0146aa4537c34077b423e6b1d7e4ddadcbdbf9cea64"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:46:53.403199Z","signature_b64":"5vooQE6kBOcMytE9NfmeWoPVOioa1Z4SK9wtCzg3ysPSWqDKQPSNjEJmNFEpLlbqCVOizbUzuHA0EVQ1/RF5AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a63a4644bcee72c13c22a42404aa6e740aa390bf7e9c0b78d920e7f0ed556ecf","last_reissued_at":"2026-07-05T10:46:53.402597Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:46:53.402597Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Rethinking Patch Dependence for Masked Autoencoders","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Adam Yala, Alexei A. Efros, Baifeng Shi, Ken Goldberg, Letian Fu, Long Lian, Renhao Wang, Trevor Darrell, Xudong Wang","submitted_at":"2024-01-25T18:49:57Z","abstract_excerpt":"In this work, we examine the impact of inter-patch dependencies in the decoder of masked autoencoders (MAE) on representation learning. We decompose the decoding mechanism for masked reconstruction into self-attention between mask tokens and cross-attention between masked and visible tokens. Our findings reveal that MAE reconstructs coherent images from visible patches not through interactions between patches in the decoder but by learning a global representation within the encoder. This discovery leads us to propose a simple visual pretraining framework: cross-attention masked autoencoders (C"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.14391","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/2401.14391/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":"2401.14391","created_at":"2026-07-05T10:46:53.402658+00:00"},{"alias_kind":"arxiv_version","alias_value":"2401.14391v2","created_at":"2026-07-05T10:46:53.402658+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.14391","created_at":"2026-07-05T10:46:53.402658+00:00"},{"alias_kind":"pith_short_12","alias_value":"UY5EMRF45ZZM","created_at":"2026-07-05T10:46:53.402658+00:00"},{"alias_kind":"pith_short_16","alias_value":"UY5EMRF45ZZMCPBC","created_at":"2026-07-05T10:46:53.402658+00:00"},{"alias_kind":"pith_short_8","alias_value":"UY5EMRF4","created_at":"2026-07-05T10:46:53.402658+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.31108","citing_title":"Remembering by Reconstructing: Domain Incremental Learning With Test-Time Training on Video Streams","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2512.03637","citing_title":"AaSP: Aliasing-aware Self-Supervised Pre-Training for Audio Spectrogram Transformers","ref_index":36,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UY5EMRF45ZZMCPBCUQSAJKTOOQ","json":"https://pith.science/pith/UY5EMRF45ZZMCPBCUQSAJKTOOQ.json","graph_json":"https://pith.science/api/pith-number/UY5EMRF45ZZMCPBCUQSAJKTOOQ/graph.json","events_json":"https://pith.science/api/pith-number/UY5EMRF45ZZMCPBCUQSAJKTOOQ/events.json","paper":"https://pith.science/paper/UY5EMRF4"},"agent_actions":{"view_html":"https://pith.science/pith/UY5EMRF45ZZMCPBCUQSAJKTOOQ","download_json":"https://pith.science/pith/UY5EMRF45ZZMCPBCUQSAJKTOOQ.json","view_paper":"https://pith.science/paper/UY5EMRF4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2401.14391&json=true","fetch_graph":"https://pith.science/api/pith-number/UY5EMRF45ZZMCPBCUQSAJKTOOQ/graph.json","fetch_events":"https://pith.science/api/pith-number/UY5EMRF45ZZMCPBCUQSAJKTOOQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UY5EMRF45ZZMCPBCUQSAJKTOOQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UY5EMRF45ZZMCPBCUQSAJKTOOQ/action/storage_attestation","attest_author":"https://pith.science/pith/UY5EMRF45ZZMCPBCUQSAJKTOOQ/action/author_attestation","sign_citation":"https://pith.science/pith/UY5EMRF45ZZMCPBCUQSAJKTOOQ/action/citation_signature","submit_replication":"https://pith.science/pith/UY5EMRF45ZZMCPBCUQSAJKTOOQ/action/replication_record"}},"created_at":"2026-07-05T10:46:53.402658+00:00","updated_at":"2026-07-05T10:46:53.402658+00:00"}