{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:WKXUJ6YJ4YBOM4ISNXJR554EFM","short_pith_number":"pith:WKXUJ6YJ","schema_version":"1.0","canonical_sha256":"b2af44fb09e602e671126dd31ef7842b026cb2661d154c48ff615e361f9c23d4","source":{"kind":"arxiv","id":"2106.07617","version":4},"attestation_state":"computed","paper":{"title":"Delving Deep into the Generalization of Vision Transformers under Distribution Shifts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Chongzhi Zhang, Daisheng Jin, Haiyu Zhao, Mingyuan Zhang, Qiang Zhou, Shanghang Zhang, Xianglong Liu, Zhongang Cai, Ziwei Liu","submitted_at":"2021-06-14T17:21:41Z","abstract_excerpt":"Vision Transformers (ViTs) have achieved impressive performance on various vision tasks, yet their generalization under distribution shifts (DS) is rarely understood. In this work, we comprehensively study the out-of-distribution (OOD) generalization of ViTs. For systematic investigation, we first present a taxonomy of DS. We then perform extensive evaluations of ViT variants under different DS and compare their generalization with Convolutional Neural Network (CNN) models. Important observations are obtained: 1) ViTs learn weaker biases on backgrounds and textures, while they are equipped wit"},"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":"2106.07617","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2021-06-14T17:21:41Z","cross_cats_sorted":[],"title_canon_sha256":"f90c61706aeebfc51a06ba39cff084e214025d67eee08678aaf95ea01554013e","abstract_canon_sha256":"fa585d62c05f34c05e6b2b30bc5fd3ea4308acd9c4c1d281ee942ecf9e5e9eaa"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:02:39.130651Z","signature_b64":"gyXxD/lY65DwSc9eDnDyGFYh+UpMrPEpNY8PKjUKgzjuTLGrIBvK4ZITU+WZUmhWr77rMrdnUKO/gOOxzNMpCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b2af44fb09e602e671126dd31ef7842b026cb2661d154c48ff615e361f9c23d4","last_reissued_at":"2026-07-05T04:02:39.129745Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:02:39.129745Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Delving Deep into the Generalization of Vision Transformers under Distribution Shifts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Chongzhi Zhang, Daisheng Jin, Haiyu Zhao, Mingyuan Zhang, Qiang Zhou, Shanghang Zhang, Xianglong Liu, Zhongang Cai, Ziwei Liu","submitted_at":"2021-06-14T17:21:41Z","abstract_excerpt":"Vision Transformers (ViTs) have achieved impressive performance on various vision tasks, yet their generalization under distribution shifts (DS) is rarely understood. In this work, we comprehensively study the out-of-distribution (OOD) generalization of ViTs. For systematic investigation, we first present a taxonomy of DS. We then perform extensive evaluations of ViT variants under different DS and compare their generalization with Convolutional Neural Network (CNN) models. Important observations are obtained: 1) ViTs learn weaker biases on backgrounds and textures, while they are equipped wit"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2106.07617","kind":"arxiv","version":4},"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/2106.07617/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":"2106.07617","created_at":"2026-07-05T04:02:39.129805+00:00"},{"alias_kind":"arxiv_version","alias_value":"2106.07617v4","created_at":"2026-07-05T04:02:39.129805+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2106.07617","created_at":"2026-07-05T04:02:39.129805+00:00"},{"alias_kind":"pith_short_12","alias_value":"WKXUJ6YJ4YBO","created_at":"2026-07-05T04:02:39.129805+00:00"},{"alias_kind":"pith_short_16","alias_value":"WKXUJ6YJ4YBOM4IS","created_at":"2026-07-05T04:02:39.129805+00:00"},{"alias_kind":"pith_short_8","alias_value":"WKXUJ6YJ","created_at":"2026-07-05T04:02:39.129805+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/WKXUJ6YJ4YBOM4ISNXJR554EFM","json":"https://pith.science/pith/WKXUJ6YJ4YBOM4ISNXJR554EFM.json","graph_json":"https://pith.science/api/pith-number/WKXUJ6YJ4YBOM4ISNXJR554EFM/graph.json","events_json":"https://pith.science/api/pith-number/WKXUJ6YJ4YBOM4ISNXJR554EFM/events.json","paper":"https://pith.science/paper/WKXUJ6YJ"},"agent_actions":{"view_html":"https://pith.science/pith/WKXUJ6YJ4YBOM4ISNXJR554EFM","download_json":"https://pith.science/pith/WKXUJ6YJ4YBOM4ISNXJR554EFM.json","view_paper":"https://pith.science/paper/WKXUJ6YJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2106.07617&json=true","fetch_graph":"https://pith.science/api/pith-number/WKXUJ6YJ4YBOM4ISNXJR554EFM/graph.json","fetch_events":"https://pith.science/api/pith-number/WKXUJ6YJ4YBOM4ISNXJR554EFM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WKXUJ6YJ4YBOM4ISNXJR554EFM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WKXUJ6YJ4YBOM4ISNXJR554EFM/action/storage_attestation","attest_author":"https://pith.science/pith/WKXUJ6YJ4YBOM4ISNXJR554EFM/action/author_attestation","sign_citation":"https://pith.science/pith/WKXUJ6YJ4YBOM4ISNXJR554EFM/action/citation_signature","submit_replication":"https://pith.science/pith/WKXUJ6YJ4YBOM4ISNXJR554EFM/action/replication_record"}},"created_at":"2026-07-05T04:02:39.129805+00:00","updated_at":"2026-07-05T04:02:39.129805+00:00"}