{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:57OKZUAEQDIOUZXUD63RFAERQV","short_pith_number":"pith:57OKZUAE","schema_version":"1.0","canonical_sha256":"efdcacd00480d0ea66f41fb7128091856cdbe8d572b0a43af9803788f7d0d797","source":{"kind":"arxiv","id":"2402.02379","version":2},"attestation_state":"computed","paper":{"title":"Unveiling the Deficiencies of Pre-trained Text-and-Layout Models in Real-world Visually-rich Document Information Extraction","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chenshu Yuan, Chong Zhang, Mingxu Chai, Qi Zhang, Ya Guo, Yi Tu, Yixi Zhao, Yue Zhang, Yulu Xie, Ziyu Shen","submitted_at":"2024-02-04T07:33:45Z","abstract_excerpt":"Recently developed pre-trained text-and-layout models (PTLMs) have shown remarkable success in multiple information extraction tasks on visually-rich documents (VrDs). However, despite achieving extremely high performance on benchmarks, their real-world performance falls short of expectations. Owing to this issue, we investigate the prevailing evaluation pipeline to reveal that: (1) The inadequate annotations within benchmark datasets introduce spurious correlations between task inputs and labels, which would lead to overly-optimistic estimation of model performance. (2) The evaluation solely "},"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":"2402.02379","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-02-04T07:33:45Z","cross_cats_sorted":[],"title_canon_sha256":"6e4e88f847f3ae8076768afdaa70b3a7b5d607b2afce3e8350153b15bdc588f9","abstract_canon_sha256":"b3d3417bd63b4af022ba6940c6c22462debc08da0f511ebafe532250f419a5d4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:48:14.448769Z","signature_b64":"YqXZjQrFADDaAObIrDvTKtsDmkFz+4kwZEI/+wqEwkd8A/OJOolsxJ52G0zVS6sunU7TEbXJDlqmW4Vx/Ou+DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"efdcacd00480d0ea66f41fb7128091856cdbe8d572b0a43af9803788f7d0d797","last_reissued_at":"2026-07-05T10:48:14.448260Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:48:14.448260Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Unveiling the Deficiencies of Pre-trained Text-and-Layout Models in Real-world Visually-rich Document Information Extraction","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chenshu Yuan, Chong Zhang, Mingxu Chai, Qi Zhang, Ya Guo, Yi Tu, Yixi Zhao, Yue Zhang, Yulu Xie, Ziyu Shen","submitted_at":"2024-02-04T07:33:45Z","abstract_excerpt":"Recently developed pre-trained text-and-layout models (PTLMs) have shown remarkable success in multiple information extraction tasks on visually-rich documents (VrDs). However, despite achieving extremely high performance on benchmarks, their real-world performance falls short of expectations. Owing to this issue, we investigate the prevailing evaluation pipeline to reveal that: (1) The inadequate annotations within benchmark datasets introduce spurious correlations between task inputs and labels, which would lead to overly-optimistic estimation of model performance. (2) The evaluation solely "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.02379","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/2402.02379/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":"2402.02379","created_at":"2026-07-05T10:48:14.448322+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.02379v2","created_at":"2026-07-05T10:48:14.448322+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.02379","created_at":"2026-07-05T10:48:14.448322+00:00"},{"alias_kind":"pith_short_12","alias_value":"57OKZUAEQDIO","created_at":"2026-07-05T10:48:14.448322+00:00"},{"alias_kind":"pith_short_16","alias_value":"57OKZUAEQDIOUZXU","created_at":"2026-07-05T10:48:14.448322+00:00"},{"alias_kind":"pith_short_8","alias_value":"57OKZUAE","created_at":"2026-07-05T10:48:14.448322+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.25927","citing_title":"Information Extraction from Electricity Invoices with General-Purpose Large Language Models","ref_index":27,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/57OKZUAEQDIOUZXUD63RFAERQV","json":"https://pith.science/pith/57OKZUAEQDIOUZXUD63RFAERQV.json","graph_json":"https://pith.science/api/pith-number/57OKZUAEQDIOUZXUD63RFAERQV/graph.json","events_json":"https://pith.science/api/pith-number/57OKZUAEQDIOUZXUD63RFAERQV/events.json","paper":"https://pith.science/paper/57OKZUAE"},"agent_actions":{"view_html":"https://pith.science/pith/57OKZUAEQDIOUZXUD63RFAERQV","download_json":"https://pith.science/pith/57OKZUAEQDIOUZXUD63RFAERQV.json","view_paper":"https://pith.science/paper/57OKZUAE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.02379&json=true","fetch_graph":"https://pith.science/api/pith-number/57OKZUAEQDIOUZXUD63RFAERQV/graph.json","fetch_events":"https://pith.science/api/pith-number/57OKZUAEQDIOUZXUD63RFAERQV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/57OKZUAEQDIOUZXUD63RFAERQV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/57OKZUAEQDIOUZXUD63RFAERQV/action/storage_attestation","attest_author":"https://pith.science/pith/57OKZUAEQDIOUZXUD63RFAERQV/action/author_attestation","sign_citation":"https://pith.science/pith/57OKZUAEQDIOUZXUD63RFAERQV/action/citation_signature","submit_replication":"https://pith.science/pith/57OKZUAEQDIOUZXUD63RFAERQV/action/replication_record"}},"created_at":"2026-07-05T10:48:14.448322+00:00","updated_at":"2026-07-05T10:48:14.448322+00:00"}