{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:KWM2AYUTZDHFV57RHBNK55J57H","short_pith_number":"pith:KWM2AYUT","schema_version":"1.0","canonical_sha256":"5599a06293c8ce5af7f1385aaef53df9d3cc347aa207538805687a9cd0d510b6","source":{"kind":"arxiv","id":"2103.14586","version":2},"attestation_state":"computed","paper":{"title":"Understanding Robustness of Transformers for Image Classification","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Andreas Veit, Ayan Chakrabarti, Daliang Li, Daniel Glasner, Srinadh Bhojanapalli, Thomas Unterthiner","submitted_at":"2021-03-26T16:47:55Z","abstract_excerpt":"Deep Convolutional Neural Networks (CNNs) have long been the architecture of choice for computer vision tasks. Recently, Transformer-based architectures like Vision Transformer (ViT) have matched or even surpassed ResNets for image classification. However, details of the Transformer architecture -- such as the use of non-overlapping patches -- lead one to wonder whether these networks are as robust. In this paper, we perform an extensive study of a variety of different measures of robustness of ViT models and compare the findings to ResNet baselines. We investigate robustness to input perturba"},"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":"2103.14586","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2021-03-26T16:47:55Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"2a0bae6349068ae1ff861dabcf336b138c1e474758d3e7f376bf9536db43143d","abstract_canon_sha256":"51a54ee52eca7354b69ecfe57f8a7c605f23a5bc11162146b36efd8157dac525"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:20:54.830900Z","signature_b64":"YNW1H0n2cDCU7+N4ox80UYUxmlfeU/SC2dcM6At492m/GMXNnHVtdIwyrktvdyok9AfYJmnYqcDgwgnprd6BAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5599a06293c8ce5af7f1385aaef53df9d3cc347aa207538805687a9cd0d510b6","last_reissued_at":"2026-07-05T03:20:54.830477Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:20:54.830477Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Understanding Robustness of Transformers for Image Classification","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Andreas Veit, Ayan Chakrabarti, Daliang Li, Daniel Glasner, Srinadh Bhojanapalli, Thomas Unterthiner","submitted_at":"2021-03-26T16:47:55Z","abstract_excerpt":"Deep Convolutional Neural Networks (CNNs) have long been the architecture of choice for computer vision tasks. Recently, Transformer-based architectures like Vision Transformer (ViT) have matched or even surpassed ResNets for image classification. However, details of the Transformer architecture -- such as the use of non-overlapping patches -- lead one to wonder whether these networks are as robust. In this paper, we perform an extensive study of a variety of different measures of robustness of ViT models and compare the findings to ResNet baselines. We investigate robustness to input perturba"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2103.14586","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/2103.14586/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":"2103.14586","created_at":"2026-07-05T03:20:54.830537+00:00"},{"alias_kind":"arxiv_version","alias_value":"2103.14586v2","created_at":"2026-07-05T03:20:54.830537+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2103.14586","created_at":"2026-07-05T03:20:54.830537+00:00"},{"alias_kind":"pith_short_12","alias_value":"KWM2AYUTZDHF","created_at":"2026-07-05T03:20:54.830537+00:00"},{"alias_kind":"pith_short_16","alias_value":"KWM2AYUTZDHFV57R","created_at":"2026-07-05T03:20:54.830537+00:00"},{"alias_kind":"pith_short_8","alias_value":"KWM2AYUT","created_at":"2026-07-05T03:20:54.830537+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/KWM2AYUTZDHFV57RHBNK55J57H","json":"https://pith.science/pith/KWM2AYUTZDHFV57RHBNK55J57H.json","graph_json":"https://pith.science/api/pith-number/KWM2AYUTZDHFV57RHBNK55J57H/graph.json","events_json":"https://pith.science/api/pith-number/KWM2AYUTZDHFV57RHBNK55J57H/events.json","paper":"https://pith.science/paper/KWM2AYUT"},"agent_actions":{"view_html":"https://pith.science/pith/KWM2AYUTZDHFV57RHBNK55J57H","download_json":"https://pith.science/pith/KWM2AYUTZDHFV57RHBNK55J57H.json","view_paper":"https://pith.science/paper/KWM2AYUT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2103.14586&json=true","fetch_graph":"https://pith.science/api/pith-number/KWM2AYUTZDHFV57RHBNK55J57H/graph.json","fetch_events":"https://pith.science/api/pith-number/KWM2AYUTZDHFV57RHBNK55J57H/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KWM2AYUTZDHFV57RHBNK55J57H/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KWM2AYUTZDHFV57RHBNK55J57H/action/storage_attestation","attest_author":"https://pith.science/pith/KWM2AYUTZDHFV57RHBNK55J57H/action/author_attestation","sign_citation":"https://pith.science/pith/KWM2AYUTZDHFV57RHBNK55J57H/action/citation_signature","submit_replication":"https://pith.science/pith/KWM2AYUTZDHFV57RHBNK55J57H/action/replication_record"}},"created_at":"2026-07-05T03:20:54.830537+00:00","updated_at":"2026-07-05T03:20:54.830537+00:00"}