{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:U5G67EGIYSYW5253IWMGPHQXMB","short_pith_number":"pith:U5G67EGI","schema_version":"1.0","canonical_sha256":"a74def90c8c4b16eebbb4598679e1760546fa53071ea3fdcc31fe50692def53f","source":{"kind":"arxiv","id":"2111.10493","version":2},"attestation_state":"computed","paper":{"title":"Discrete Representations Strengthen Vision Transformer Robustness","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Carl Vondrick, Chengzhi Mao, Irfan Essa, Lu Jiang, Mostafa Dehghani, Rahul Sukthankar","submitted_at":"2021-11-20T01:49:56Z","abstract_excerpt":"Vision Transformer (ViT) is emerging as the state-of-the-art architecture for image recognition. While recent studies suggest that ViTs are more robust than their convolutional counterparts, our experiments find that ViTs trained on ImageNet are overly reliant on local textures and fail to make adequate use of shape information. ViTs thus have difficulties generalizing to out-of-distribution, real-world data. To address this deficiency, we present a simple and effective architecture modification to ViT's input layer by adding discrete tokens produced by a vector-quantized encoder. Different fr"},"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":"2111.10493","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2021-11-20T01:49:56Z","cross_cats_sorted":[],"title_canon_sha256":"4eb266f70b9d870d0e13689fda46ca1c9d4136727ceed6c84e3830185d06f653","abstract_canon_sha256":"3fd4c598334ccfda2c59e968f3856d6233032d6a271254d62effb82571ccd31d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:10:51.872267Z","signature_b64":"GWJXl98LSAr9mqZyWFjAE55QMPaGNB0g/tOsEEZfzYFVlIjvMq0Ad0nwZgPdYi7sK6QjgPzUfUBdgyqfEPptCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a74def90c8c4b16eebbb4598679e1760546fa53071ea3fdcc31fe50692def53f","last_reissued_at":"2026-07-05T04:10:51.871792Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:10:51.871792Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Discrete Representations Strengthen Vision Transformer Robustness","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Carl Vondrick, Chengzhi Mao, Irfan Essa, Lu Jiang, Mostafa Dehghani, Rahul Sukthankar","submitted_at":"2021-11-20T01:49:56Z","abstract_excerpt":"Vision Transformer (ViT) is emerging as the state-of-the-art architecture for image recognition. While recent studies suggest that ViTs are more robust than their convolutional counterparts, our experiments find that ViTs trained on ImageNet are overly reliant on local textures and fail to make adequate use of shape information. ViTs thus have difficulties generalizing to out-of-distribution, real-world data. To address this deficiency, we present a simple and effective architecture modification to ViT's input layer by adding discrete tokens produced by a vector-quantized encoder. Different fr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2111.10493","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/2111.10493/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":"2111.10493","created_at":"2026-07-05T04:10:51.871845+00:00"},{"alias_kind":"arxiv_version","alias_value":"2111.10493v2","created_at":"2026-07-05T04:10:51.871845+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2111.10493","created_at":"2026-07-05T04:10:51.871845+00:00"},{"alias_kind":"pith_short_12","alias_value":"U5G67EGIYSYW","created_at":"2026-07-05T04:10:51.871845+00:00"},{"alias_kind":"pith_short_16","alias_value":"U5G67EGIYSYW5253","created_at":"2026-07-05T04:10:51.871845+00:00"},{"alias_kind":"pith_short_8","alias_value":"U5G67EGI","created_at":"2026-07-05T04:10:51.871845+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.05855","citing_title":"EEGDancer: Dynamic Emotion Latent Space Masked Modeling with Reinforcement Learning for EEG Continuous Emotion Prediction","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2605.13517","citing_title":"ArcVQ-VAE: A Spherical Vector Quantization Framework with ArcCosine Additive Margin","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2605.13517","citing_title":"ArcVQ-VAE: A Spherical Vector Quantization Framework with ArcCosine Additive Margin","ref_index":6,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/U5G67EGIYSYW5253IWMGPHQXMB","json":"https://pith.science/pith/U5G67EGIYSYW5253IWMGPHQXMB.json","graph_json":"https://pith.science/api/pith-number/U5G67EGIYSYW5253IWMGPHQXMB/graph.json","events_json":"https://pith.science/api/pith-number/U5G67EGIYSYW5253IWMGPHQXMB/events.json","paper":"https://pith.science/paper/U5G67EGI"},"agent_actions":{"view_html":"https://pith.science/pith/U5G67EGIYSYW5253IWMGPHQXMB","download_json":"https://pith.science/pith/U5G67EGIYSYW5253IWMGPHQXMB.json","view_paper":"https://pith.science/paper/U5G67EGI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2111.10493&json=true","fetch_graph":"https://pith.science/api/pith-number/U5G67EGIYSYW5253IWMGPHQXMB/graph.json","fetch_events":"https://pith.science/api/pith-number/U5G67EGIYSYW5253IWMGPHQXMB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/U5G67EGIYSYW5253IWMGPHQXMB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/U5G67EGIYSYW5253IWMGPHQXMB/action/storage_attestation","attest_author":"https://pith.science/pith/U5G67EGIYSYW5253IWMGPHQXMB/action/author_attestation","sign_citation":"https://pith.science/pith/U5G67EGIYSYW5253IWMGPHQXMB/action/citation_signature","submit_replication":"https://pith.science/pith/U5G67EGIYSYW5253IWMGPHQXMB/action/replication_record"}},"created_at":"2026-07-05T04:10:51.871845+00:00","updated_at":"2026-07-05T04:10:51.871845+00:00"}