{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:Y3M33WJOACZRREIUQSWPQV2F5L","short_pith_number":"pith:Y3M33WJO","schema_version":"1.0","canonical_sha256":"c6d9bdd92e00b318911484acf85745ead858f7ddf786fa02b3d664fefc3b723a","source":{"kind":"arxiv","id":"2105.02723","version":1},"attestation_state":"computed","paper":{"title":"Do You Even Need Attention? A Stack of Feed-Forward Layers Does Surprisingly Well on ImageNet","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Luke Melas-Kyriazi","submitted_at":"2021-05-06T14:42:39Z","abstract_excerpt":"The strong performance of vision transformers on image classification and other vision tasks is often attributed to the design of their multi-head attention layers. However, the extent to which attention is responsible for this strong performance remains unclear. In this short report, we ask: is the attention layer even necessary? Specifically, we replace the attention layer in a vision transformer with a feed-forward layer applied over the patch dimension. The resulting architecture is simply a series of feed-forward layers applied over the patch and feature dimensions in an alternating fashi"},"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":"2105.02723","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2021-05-06T14:42:39Z","cross_cats_sorted":[],"title_canon_sha256":"2a0dd56b352c6e38b21c0311bffba37bc2e862564f999c73795e4133cc6200b6","abstract_canon_sha256":"26cb3b92f5c2b2a9ad3f5a740e597083120b82f289e68f6a4cf0e9488825c69a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:38:09.803147Z","signature_b64":"YPE187BCWhWMtmbM5Y79WelCpH/UDxT2KPZ4EzWmWb8U2/4zqiYoSwCKUzEr6VyfR6R7EPKikuP7N7uitjUHDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c6d9bdd92e00b318911484acf85745ead858f7ddf786fa02b3d664fefc3b723a","last_reissued_at":"2026-07-05T02:38:09.802682Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:38:09.802682Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Do You Even Need Attention? A Stack of Feed-Forward Layers Does Surprisingly Well on ImageNet","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Luke Melas-Kyriazi","submitted_at":"2021-05-06T14:42:39Z","abstract_excerpt":"The strong performance of vision transformers on image classification and other vision tasks is often attributed to the design of their multi-head attention layers. However, the extent to which attention is responsible for this strong performance remains unclear. In this short report, we ask: is the attention layer even necessary? Specifically, we replace the attention layer in a vision transformer with a feed-forward layer applied over the patch dimension. The resulting architecture is simply a series of feed-forward layers applied over the patch and feature dimensions in an alternating fashi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2105.02723","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/2105.02723/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":"2105.02723","created_at":"2026-07-05T02:38:09.802738+00:00"},{"alias_kind":"arxiv_version","alias_value":"2105.02723v1","created_at":"2026-07-05T02:38:09.802738+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2105.02723","created_at":"2026-07-05T02:38:09.802738+00:00"},{"alias_kind":"pith_short_12","alias_value":"Y3M33WJOACZR","created_at":"2026-07-05T02:38:09.802738+00:00"},{"alias_kind":"pith_short_16","alias_value":"Y3M33WJOACZRREIU","created_at":"2026-07-05T02:38:09.802738+00:00"},{"alias_kind":"pith_short_8","alias_value":"Y3M33WJO","created_at":"2026-07-05T02:38:09.802738+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.06949","citing_title":"SpiS-GAN: Spiral-Modulated Handwriting Synthesis with Star Operation","ref_index":33,"is_internal_anchor":true},{"citing_arxiv_id":"2605.06683","citing_title":"Toeplitz MLP Mixers are Low Complexity, Information-Rich Sequence Models","ref_index":58,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Y3M33WJOACZRREIUQSWPQV2F5L","json":"https://pith.science/pith/Y3M33WJOACZRREIUQSWPQV2F5L.json","graph_json":"https://pith.science/api/pith-number/Y3M33WJOACZRREIUQSWPQV2F5L/graph.json","events_json":"https://pith.science/api/pith-number/Y3M33WJOACZRREIUQSWPQV2F5L/events.json","paper":"https://pith.science/paper/Y3M33WJO"},"agent_actions":{"view_html":"https://pith.science/pith/Y3M33WJOACZRREIUQSWPQV2F5L","download_json":"https://pith.science/pith/Y3M33WJOACZRREIUQSWPQV2F5L.json","view_paper":"https://pith.science/paper/Y3M33WJO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2105.02723&json=true","fetch_graph":"https://pith.science/api/pith-number/Y3M33WJOACZRREIUQSWPQV2F5L/graph.json","fetch_events":"https://pith.science/api/pith-number/Y3M33WJOACZRREIUQSWPQV2F5L/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Y3M33WJOACZRREIUQSWPQV2F5L/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Y3M33WJOACZRREIUQSWPQV2F5L/action/storage_attestation","attest_author":"https://pith.science/pith/Y3M33WJOACZRREIUQSWPQV2F5L/action/author_attestation","sign_citation":"https://pith.science/pith/Y3M33WJOACZRREIUQSWPQV2F5L/action/citation_signature","submit_replication":"https://pith.science/pith/Y3M33WJOACZRREIUQSWPQV2F5L/action/replication_record"}},"created_at":"2026-07-05T02:38:09.802738+00:00","updated_at":"2026-07-05T02:38:09.802738+00:00"}