{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:VXS2PN65FYPPRIKZBATMMY5OEH","short_pith_number":"pith:VXS2PN65","schema_version":"1.0","canonical_sha256":"ade5a7b7dd2e1ef8a1590826c663ae21c1781eb86d73db881f31a2d416b1269e","source":{"kind":"arxiv","id":"2111.15664","version":5},"attestation_state":"computed","paper":{"title":"OCR-free Document Understanding Transformer","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Dongyoon Han, Geewook Kim, Jeongyeon Nam, Jinyeong Yim, Jinyoung Park, Moonbin Yim, Sangdoo Yun, Seunghyun Park, Teakgyu Hong, Wonseok Hwang","submitted_at":"2021-11-30T18:55:19Z","abstract_excerpt":"Understanding document images (e.g., invoices) is a core but challenging task since it requires complex functions such as reading text and a holistic understanding of the document. Current Visual Document Understanding (VDU) methods outsource the task of reading text to off-the-shelf Optical Character Recognition (OCR) engines and focus on the understanding task with the OCR outputs. Although such OCR-based approaches have shown promising performance, they suffer from 1) high computational costs for using OCR; 2) inflexibility of OCR models on languages or types of document; 3) OCR error propa"},"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":"2111.15664","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-11-30T18:55:19Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"8b6320ce9a458a9a71aa2db0f360c803b6cbc4f1ee12324aa996e41cbb140d6c","abstract_canon_sha256":"15c4c8c080e88f0ba294f57261a57c6c830a06f7f4830a0fd47b1fdfb8934146"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:03:49.955650Z","signature_b64":"B3rOXBJoAxg9rX233MHNBiWCoCQUj2vpc9UHAvGOW0ggWrp7UtsmkvxDPtYJpJyYkMvWS4A0LMsI1d9nv9ujAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ade5a7b7dd2e1ef8a1590826c663ae21c1781eb86d73db881f31a2d416b1269e","last_reissued_at":"2026-07-05T05:03:49.955162Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:03:49.955162Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"OCR-free Document Understanding Transformer","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Dongyoon Han, Geewook Kim, Jeongyeon Nam, Jinyeong Yim, Jinyoung Park, Moonbin Yim, Sangdoo Yun, Seunghyun Park, Teakgyu Hong, Wonseok Hwang","submitted_at":"2021-11-30T18:55:19Z","abstract_excerpt":"Understanding document images (e.g., invoices) is a core but challenging task since it requires complex functions such as reading text and a holistic understanding of the document. Current Visual Document Understanding (VDU) methods outsource the task of reading text to off-the-shelf Optical Character Recognition (OCR) engines and focus on the understanding task with the OCR outputs. Although such OCR-based approaches have shown promising performance, they suffer from 1) high computational costs for using OCR; 2) inflexibility of OCR models on languages or types of document; 3) OCR error propa"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2111.15664","kind":"arxiv","version":5},"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/2111.15664/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":"2111.15664","created_at":"2026-07-05T05:03:49.955221+00:00"},{"alias_kind":"arxiv_version","alias_value":"2111.15664v5","created_at":"2026-07-05T05:03:49.955221+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2111.15664","created_at":"2026-07-05T05:03:49.955221+00:00"},{"alias_kind":"pith_short_12","alias_value":"VXS2PN65FYPP","created_at":"2026-07-05T05:03:49.955221+00:00"},{"alias_kind":"pith_short_16","alias_value":"VXS2PN65FYPPRIKZ","created_at":"2026-07-05T05:03:49.955221+00:00"},{"alias_kind":"pith_short_8","alias_value":"VXS2PN65","created_at":"2026-07-05T05:03:49.955221+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":14,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.24302","citing_title":"TrOCR for Medieval HTR: A Systematic Ablation Study with Cross-Dataset Validation","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2606.02162","citing_title":"Multimodal Approaches for Visually-Rich Document Type Classification: A Comparative Analysis","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27446","citing_title":"Causal Connections: Leveraging Multilingual Fine-Tuning for Financial QA@FinCausal 2026","ref_index":60,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27316","citing_title":"LLM-Based Examination of Eligibility Criteria from Securities Prospectuses at the German Central Bank","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2502.20295","citing_title":"Judge a Book by its Cover: Investigating Multi-Modal LLMs for Multi-Page Handwritten Document Transcription","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17447","citing_title":"FastOCR: Dynamic Visual Fixation via KV Cache Pruning for Efficient Document Parsing","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17159","citing_title":"MADP: A Multi-Agent Pipeline for Sustainable Document Processing with Human-in-the-Loop","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19866","citing_title":"Structured Layout Priors for Robust Out-of-Distribution Visual Document Understanding","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2308.13418","citing_title":"Nougat: Neural Optical Understanding for Academic Documents","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2602.01785","citing_title":"CodeOCR: On the Effectiveness of Vision Language Models in Code Understanding","ref_index":52,"is_internal_anchor":false},{"citing_arxiv_id":"2603.19790","citing_title":"From Plausibility to Verifiability: Risk-Controlled Generative OCR with Vision-Language Models","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2604.03476","citing_title":"Fine-tuning DeepSeek-OCR-2 for Molecular Structure Recognition","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16504","citing_title":"From Handwriting to Structured Data: Benchmarking AI Digitisation of Handwritten Forms","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04771","citing_title":"MinerU2.5-Pro: Pushing the Limits of Data-Centric Document Parsing at Scale","ref_index":16,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VXS2PN65FYPPRIKZBATMMY5OEH","json":"https://pith.science/pith/VXS2PN65FYPPRIKZBATMMY5OEH.json","graph_json":"https://pith.science/api/pith-number/VXS2PN65FYPPRIKZBATMMY5OEH/graph.json","events_json":"https://pith.science/api/pith-number/VXS2PN65FYPPRIKZBATMMY5OEH/events.json","paper":"https://pith.science/paper/VXS2PN65"},"agent_actions":{"view_html":"https://pith.science/pith/VXS2PN65FYPPRIKZBATMMY5OEH","download_json":"https://pith.science/pith/VXS2PN65FYPPRIKZBATMMY5OEH.json","view_paper":"https://pith.science/paper/VXS2PN65","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2111.15664&json=true","fetch_graph":"https://pith.science/api/pith-number/VXS2PN65FYPPRIKZBATMMY5OEH/graph.json","fetch_events":"https://pith.science/api/pith-number/VXS2PN65FYPPRIKZBATMMY5OEH/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VXS2PN65FYPPRIKZBATMMY5OEH/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VXS2PN65FYPPRIKZBATMMY5OEH/action/storage_attestation","attest_author":"https://pith.science/pith/VXS2PN65FYPPRIKZBATMMY5OEH/action/author_attestation","sign_citation":"https://pith.science/pith/VXS2PN65FYPPRIKZBATMMY5OEH/action/citation_signature","submit_replication":"https://pith.science/pith/VXS2PN65FYPPRIKZBATMMY5OEH/action/replication_record"}},"created_at":"2026-07-05T05:03:49.955221+00:00","updated_at":"2026-07-05T05:03:49.955221+00:00"}