{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:H4Q3CRZLLJGO52LL37KJPDSRIY","short_pith_number":"pith:H4Q3CRZL","schema_version":"1.0","canonical_sha256":"3f21b1472b5a4ceee96bdfd4978e51462a4c416031cad6b529d810b4f48afbcf","source":{"kind":"arxiv","id":"2111.01353","version":2},"attestation_state":"computed","paper":{"title":"Can Vision Transformers Perform Convolution?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Cho-Jui Hsieh, Di He, Shanda Li, Xiangning Chen","submitted_at":"2021-11-02T03:30:17Z","abstract_excerpt":"Several recent studies have demonstrated that attention-based networks, such as Vision Transformer (ViT), can outperform Convolutional Neural Networks (CNNs) on several computer vision tasks without using convolutional layers. This naturally leads to the following questions: Can a self-attention layer of ViT express any convolution operation? In this work, we prove that a single ViT layer with image patches as the input can perform any convolution operation constructively, where the multi-head attention mechanism and the relative positional encoding play essential roles. We further provide a l"},"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":"2111.01353","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2021-11-02T03:30:17Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"fb32c26f190d505814f8140d6153c2a5407e1af3de799ec0211c1e3872dded70","abstract_canon_sha256":"4788e8a9563b4755378c2a640da3cd365f790b7fd13d85a1c31701544975e213"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:28:42.495017Z","signature_b64":"YjSe8sx1OCqymQdiYX7NvCrpPTQ0ups5ihBzIHiB8KOcIHQEperFRoSwB4Pu60rLyX7ldsh+3DpcRj/OIjqjAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3f21b1472b5a4ceee96bdfd4978e51462a4c416031cad6b529d810b4f48afbcf","last_reissued_at":"2026-07-05T03:28:42.494674Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:28:42.494674Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Can Vision Transformers Perform Convolution?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Cho-Jui Hsieh, Di He, Shanda Li, Xiangning Chen","submitted_at":"2021-11-02T03:30:17Z","abstract_excerpt":"Several recent studies have demonstrated that attention-based networks, such as Vision Transformer (ViT), can outperform Convolutional Neural Networks (CNNs) on several computer vision tasks without using convolutional layers. This naturally leads to the following questions: Can a self-attention layer of ViT express any convolution operation? In this work, we prove that a single ViT layer with image patches as the input can perform any convolution operation constructively, where the multi-head attention mechanism and the relative positional encoding play essential roles. We further provide a l"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2111.01353","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/2111.01353/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":"2111.01353","created_at":"2026-07-05T03:28:42.494731+00:00"},{"alias_kind":"arxiv_version","alias_value":"2111.01353v2","created_at":"2026-07-05T03:28:42.494731+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2111.01353","created_at":"2026-07-05T03:28:42.494731+00:00"},{"alias_kind":"pith_short_12","alias_value":"H4Q3CRZLLJGO","created_at":"2026-07-05T03:28:42.494731+00:00"},{"alias_kind":"pith_short_16","alias_value":"H4Q3CRZLLJGO52LL","created_at":"2026-07-05T03:28:42.494731+00:00"},{"alias_kind":"pith_short_8","alias_value":"H4Q3CRZL","created_at":"2026-07-05T03:28:42.494731+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2506.05295","citing_title":"Sample Complexity and Representation Ability of Test-time Scaling Paradigms","ref_index":49,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/H4Q3CRZLLJGO52LL37KJPDSRIY","json":"https://pith.science/pith/H4Q3CRZLLJGO52LL37KJPDSRIY.json","graph_json":"https://pith.science/api/pith-number/H4Q3CRZLLJGO52LL37KJPDSRIY/graph.json","events_json":"https://pith.science/api/pith-number/H4Q3CRZLLJGO52LL37KJPDSRIY/events.json","paper":"https://pith.science/paper/H4Q3CRZL"},"agent_actions":{"view_html":"https://pith.science/pith/H4Q3CRZLLJGO52LL37KJPDSRIY","download_json":"https://pith.science/pith/H4Q3CRZLLJGO52LL37KJPDSRIY.json","view_paper":"https://pith.science/paper/H4Q3CRZL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2111.01353&json=true","fetch_graph":"https://pith.science/api/pith-number/H4Q3CRZLLJGO52LL37KJPDSRIY/graph.json","fetch_events":"https://pith.science/api/pith-number/H4Q3CRZLLJGO52LL37KJPDSRIY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/H4Q3CRZLLJGO52LL37KJPDSRIY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/H4Q3CRZLLJGO52LL37KJPDSRIY/action/storage_attestation","attest_author":"https://pith.science/pith/H4Q3CRZLLJGO52LL37KJPDSRIY/action/author_attestation","sign_citation":"https://pith.science/pith/H4Q3CRZLLJGO52LL37KJPDSRIY/action/citation_signature","submit_replication":"https://pith.science/pith/H4Q3CRZLLJGO52LL37KJPDSRIY/action/replication_record"}},"created_at":"2026-07-05T03:28:42.494731+00:00","updated_at":"2026-07-05T03:28:42.494731+00:00"}