{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:223ZEFVPLB47PVGICQYQX4VNIT","short_pith_number":"pith:223ZEFVP","schema_version":"1.0","canonical_sha256":"d6b79216af5879f7d4c814310bf2ad44e10ab442876a4149dc31bdb4196828ea","source":{"kind":"arxiv","id":"2106.11539","version":2},"attestation_state":"computed","paper":{"title":"DocFormer: End-to-End Transformer for Document Understanding","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bhargava Urala Kota, Bhavan Jasani, R. Manmatha, Srikar Appalaraju, Yusheng Xie","submitted_at":"2021-06-22T04:28:07Z","abstract_excerpt":"We present DocFormer -- a multi-modal transformer based architecture for the task of Visual Document Understanding (VDU). VDU is a challenging problem which aims to understand documents in their varied formats (forms, receipts etc.) and layouts. In addition, DocFormer is pre-trained in an unsupervised fashion using carefully designed tasks which encourage multi-modal interaction. DocFormer uses text, vision and spatial features and combines them using a novel multi-modal self-attention layer. DocFormer also shares learned spatial embeddings across modalities which makes it easy for the model t"},"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":"2106.11539","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2021-06-22T04:28:07Z","cross_cats_sorted":[],"title_canon_sha256":"7cc3bcb6e3d547115ff772b55d65da902f3027d266c106f2f1400170dd9a8033","abstract_canon_sha256":"4d6dcc6b2c0d0cde64d1b4a0483c56391da551a57bbb0057ebbcb5ec1a2c20d8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:15:34.845044Z","signature_b64":"sm0H8Pm145DOGV4dxtehwobOBldpOqt0WMBHHwifbbEUkhYsxN4aCCbiPqmIvIWPQvrdTV46KMscFI6D++hFBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d6b79216af5879f7d4c814310bf2ad44e10ab442876a4149dc31bdb4196828ea","last_reissued_at":"2026-07-05T03:15:34.844623Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:15:34.844623Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DocFormer: End-to-End Transformer for Document Understanding","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bhargava Urala Kota, Bhavan Jasani, R. Manmatha, Srikar Appalaraju, Yusheng Xie","submitted_at":"2021-06-22T04:28:07Z","abstract_excerpt":"We present DocFormer -- a multi-modal transformer based architecture for the task of Visual Document Understanding (VDU). VDU is a challenging problem which aims to understand documents in their varied formats (forms, receipts etc.) and layouts. In addition, DocFormer is pre-trained in an unsupervised fashion using carefully designed tasks which encourage multi-modal interaction. DocFormer uses text, vision and spatial features and combines them using a novel multi-modal self-attention layer. DocFormer also shares learned spatial embeddings across modalities which makes it easy for the model t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2106.11539","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/2106.11539/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":"2106.11539","created_at":"2026-07-05T03:15:34.844677+00:00"},{"alias_kind":"arxiv_version","alias_value":"2106.11539v2","created_at":"2026-07-05T03:15:34.844677+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2106.11539","created_at":"2026-07-05T03:15:34.844677+00:00"},{"alias_kind":"pith_short_12","alias_value":"223ZEFVPLB47","created_at":"2026-07-05T03:15:34.844677+00:00"},{"alias_kind":"pith_short_16","alias_value":"223ZEFVPLB47PVGI","created_at":"2026-07-05T03:15:34.844677+00:00"},{"alias_kind":"pith_short_8","alias_value":"223ZEFVP","created_at":"2026-07-05T03:15:34.844677+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.28796","citing_title":"Structure-Preserving Document Translation via Multi-Stage LLM Pipeline: A Case Study in Marathi","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2308.13418","citing_title":"Nougat: Neural Optical Understanding for Academic Documents","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10341","citing_title":"PaperFit: Vision-in-the-Loop Typesetting Optimization for Scientific Documents","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11042","citing_title":"Improving Layout Representation Learning Across Inconsistently Annotated Datasets via Agentic Harmonization","ref_index":39,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/223ZEFVPLB47PVGICQYQX4VNIT","json":"https://pith.science/pith/223ZEFVPLB47PVGICQYQX4VNIT.json","graph_json":"https://pith.science/api/pith-number/223ZEFVPLB47PVGICQYQX4VNIT/graph.json","events_json":"https://pith.science/api/pith-number/223ZEFVPLB47PVGICQYQX4VNIT/events.json","paper":"https://pith.science/paper/223ZEFVP"},"agent_actions":{"view_html":"https://pith.science/pith/223ZEFVPLB47PVGICQYQX4VNIT","download_json":"https://pith.science/pith/223ZEFVPLB47PVGICQYQX4VNIT.json","view_paper":"https://pith.science/paper/223ZEFVP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2106.11539&json=true","fetch_graph":"https://pith.science/api/pith-number/223ZEFVPLB47PVGICQYQX4VNIT/graph.json","fetch_events":"https://pith.science/api/pith-number/223ZEFVPLB47PVGICQYQX4VNIT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/223ZEFVPLB47PVGICQYQX4VNIT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/223ZEFVPLB47PVGICQYQX4VNIT/action/storage_attestation","attest_author":"https://pith.science/pith/223ZEFVPLB47PVGICQYQX4VNIT/action/author_attestation","sign_citation":"https://pith.science/pith/223ZEFVPLB47PVGICQYQX4VNIT/action/citation_signature","submit_replication":"https://pith.science/pith/223ZEFVPLB47PVGICQYQX4VNIT/action/replication_record"}},"created_at":"2026-07-05T03:15:34.844677+00:00","updated_at":"2026-07-05T03:15:34.844677+00:00"}