{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:SUTFO3WK6KPFZT5LKOYW4XXDVY","short_pith_number":"pith:SUTFO3WK","schema_version":"1.0","canonical_sha256":"9526576ecaf29e5ccfab53b16e5ee3ae1eab19568a92b569a5d8658f44e52f45","source":{"kind":"arxiv","id":"2302.03751","version":1},"attestation_state":"computed","paper":{"title":"Understanding Why ViT Trains Badly on Small Datasets: An Intuitive Perspective","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Boyuan Chen, Carter Yang, Haoran Zhu","submitted_at":"2023-02-07T20:56:21Z","abstract_excerpt":"Vision transformer (ViT) is an attention neural network architecture that is shown to be effective for computer vision tasks. However, compared to ResNet-18 with a similar number of parameters, ViT has a significantly lower evaluation accuracy when trained on small datasets. To facilitate studies in related fields, we provide a visual intuition to help understand why it is the case. We first compare the performance of the two models and confirm that ViT has less accuracy than ResNet-18 when trained on small datasets. We then interpret the results by showing attention map visualization for ViT "},"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":"2302.03751","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2023-02-07T20:56:21Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"ac9c36053604ed4d77043a932a5070bb04b352b6e14dde541f818d6ebd7b5dc0","abstract_canon_sha256":"703606d9ff24ce5967c9e6b5467e7b217bb3380afda543251bbb0ab7f10758e2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:39:57.292869Z","signature_b64":"YhP01WJratMDGxnx/C8btu/aSYKSOu0eN9Ui2J6B+uBA7xyjZel7gfLQO5YxtiQfiYB6Uehl3dvPXSfub9/cDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9526576ecaf29e5ccfab53b16e5ee3ae1eab19568a92b569a5d8658f44e52f45","last_reissued_at":"2026-07-05T05:39:57.292525Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:39:57.292525Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Understanding Why ViT Trains Badly on Small Datasets: An Intuitive Perspective","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Boyuan Chen, Carter Yang, Haoran Zhu","submitted_at":"2023-02-07T20:56:21Z","abstract_excerpt":"Vision transformer (ViT) is an attention neural network architecture that is shown to be effective for computer vision tasks. However, compared to ResNet-18 with a similar number of parameters, ViT has a significantly lower evaluation accuracy when trained on small datasets. To facilitate studies in related fields, we provide a visual intuition to help understand why it is the case. We first compare the performance of the two models and confirm that ViT has less accuracy than ResNet-18 when trained on small datasets. We then interpret the results by showing attention map visualization for ViT "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2302.03751","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/2302.03751/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":"2302.03751","created_at":"2026-07-05T05:39:57.292582+00:00"},{"alias_kind":"arxiv_version","alias_value":"2302.03751v1","created_at":"2026-07-05T05:39:57.292582+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.03751","created_at":"2026-07-05T05:39:57.292582+00:00"},{"alias_kind":"pith_short_12","alias_value":"SUTFO3WK6KPF","created_at":"2026-07-05T05:39:57.292582+00:00"},{"alias_kind":"pith_short_16","alias_value":"SUTFO3WK6KPFZT5L","created_at":"2026-07-05T05:39:57.292582+00:00"},{"alias_kind":"pith_short_8","alias_value":"SUTFO3WK","created_at":"2026-07-05T05:39:57.292582+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.09928","citing_title":"Forward-Only Convolutional Neural Networks with Learnable Channel-Class Assignment","ref_index":46,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/SUTFO3WK6KPFZT5LKOYW4XXDVY","json":"https://pith.science/pith/SUTFO3WK6KPFZT5LKOYW4XXDVY.json","graph_json":"https://pith.science/api/pith-number/SUTFO3WK6KPFZT5LKOYW4XXDVY/graph.json","events_json":"https://pith.science/api/pith-number/SUTFO3WK6KPFZT5LKOYW4XXDVY/events.json","paper":"https://pith.science/paper/SUTFO3WK"},"agent_actions":{"view_html":"https://pith.science/pith/SUTFO3WK6KPFZT5LKOYW4XXDVY","download_json":"https://pith.science/pith/SUTFO3WK6KPFZT5LKOYW4XXDVY.json","view_paper":"https://pith.science/paper/SUTFO3WK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2302.03751&json=true","fetch_graph":"https://pith.science/api/pith-number/SUTFO3WK6KPFZT5LKOYW4XXDVY/graph.json","fetch_events":"https://pith.science/api/pith-number/SUTFO3WK6KPFZT5LKOYW4XXDVY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SUTFO3WK6KPFZT5LKOYW4XXDVY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SUTFO3WK6KPFZT5LKOYW4XXDVY/action/storage_attestation","attest_author":"https://pith.science/pith/SUTFO3WK6KPFZT5LKOYW4XXDVY/action/author_attestation","sign_citation":"https://pith.science/pith/SUTFO3WK6KPFZT5LKOYW4XXDVY/action/citation_signature","submit_replication":"https://pith.science/pith/SUTFO3WK6KPFZT5LKOYW4XXDVY/action/replication_record"}},"created_at":"2026-07-05T05:39:57.292582+00:00","updated_at":"2026-07-05T05:39:57.292582+00:00"}