{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:MRWBTX6ANSALCH2ALUNUDRRNUM","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"a95951fe49943af07040052cbe2b31c810d39c9c6e61cca1684157a13fe4133d","cross_cats_sorted":["cs.CV","cs.LG"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-12T18:09:38Z","title_canon_sha256":"fc124a2f6bfc9f9191983bcdf92128c854775df7e9d78a1b75184569c9a8918c"},"schema_version":"1.0","source":{"id":"2605.12623","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.12623","created_at":"2026-05-18T03:10:00Z"},{"alias_kind":"arxiv_version","alias_value":"2605.12623v1","created_at":"2026-05-18T03:10:00Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.12623","created_at":"2026-05-18T03:10:00Z"},{"alias_kind":"pith_short_12","alias_value":"MRWBTX6ANSAL","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"MRWBTX6ANSALCH2A","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"MRWBTX6A","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:6b8b49eb345a41b6f9e054f2bdd21fa1d7fd1af944b7db4e59924ca05e4badd3","target":"graph","created_at":"2026-05-18T03:10:00Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Direct Preference Optimization (DPO) using rendering-derived ground truth as positive signal achieves stable multilingual adaptation, improving both in-domain (+1.9%) and out-of-domain (+1.8%) accuracy without measurable base-language degradation, where supervised fine-tuning degrades out-of-domain performance by up to 21%."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The differential rendering of native DOCX documents and synthetic LaTeX-based generation produce precise structural annotations that accurately represent real-world document distributions across all 82 languages without introducing new biases or distribution shifts."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"DocAtlas creates multilingual document datasets across 82 languages and shows DPO with rendered ground truth improves model accuracy by 1.7-1.9% without degrading base-language performance, unlike supervised fine-tuning."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Direct preference optimization with rendering-derived ground truth improves multilingual document understanding across 82 languages without base-language degradation."}],"snapshot_sha256":"38e7ff549c4cd97f2b79d1de71759e2cb5273a209f9b54a81f96523094dc1f0f"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"de605d39aaae8914b8445ffd563e90faab484cae48c46f137c0e77f8d585e400"},"paper":{"abstract_excerpt":"Multilingual document understanding remains limited for low-resource languages due to scarce training data and model-based annotation pipelines that perpetuate existing biases. We introduce DocAtlas, a framework that constructs high-fidelity OCR datasets and benchmarks covering 82 languages and 9 evaluation tasks. Our dual pipelines, differential rendering of native DOCX documents and synthetic LaTeX-based generation for right-to-left scripts produce precise structural annotations in a unified DocTag format encoding layout, text, and component types, without learned models for core annotation.","authors_text":"Abdullah Sohail, Ahmed Heakl, Ahmed Nassar, Fahad Shahbaz Khan, Imran Razzak, Peter W. J. Staar, Rania Elbadry, Salman Khan, Youssef Mohamed","cross_cats":["cs.CV","cs.LG"],"headline":"Direct preference optimization with rendering-derived ground truth improves multilingual document understanding across 82 languages without base-language degradation.","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-12T18:09:38Z","title":"DocAtlas: Multilingual Document Understanding Across 80+ Languages"},"references":{"count":85,"internal_anchors":14,"resolved_work":85,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"FirstName LastName , title =","work_id":"d9cab501-317f-4237-9e32-b5ead5964402","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"FirstName Alpher , title =","work_id":"42297990-8783-41a1-b0fa-8ccdbf630852","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Journal of Foo , volume = 13, number = 1, pages =","work_id":"65a8b3d0-af84-4f68-87eb-101c85ab18b2","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Journal of Foo , volume = 14, number = 1, pages =","work_id":"b3089947-bd36-4a24-9199-cc535e299537","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"FirstName Alpher and FirstName Gamow , title =","work_id":"caed320b-7cdc-41ca-bb08-00fb14feec62","year":null}],"snapshot_sha256":"6c0e904e1734960766610409aa89dcf852a4115c9170d7aea2f25766027f1c43"},"source":{"id":"2605.12623","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-14T20:58:47.305197Z","id":"08b03706-e999-40f7-97fc-16d1abfacb79","model_set":{"reader":"grok-4.3"},"one_line_summary":"DocAtlas creates multilingual document datasets across 82 languages and shows DPO with rendered ground truth improves model accuracy by 1.7-1.9% without degrading base-language performance, unlike supervised fine-tuning.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Direct preference optimization with rendering-derived ground truth improves multilingual document understanding across 82 languages without base-language degradation.","strongest_claim":"Direct Preference Optimization (DPO) using rendering-derived ground truth as positive signal achieves stable multilingual adaptation, improving both in-domain (+1.9%) and out-of-domain (+1.8%) accuracy without measurable base-language degradation, where supervised fine-tuning degrades out-of-domain performance by up to 21%.","weakest_assumption":"The differential rendering of native DOCX documents and synthetic LaTeX-based generation produce precise structural annotations that accurately represent real-world document distributions across all 82 languages without introducing new biases or distribution shifts."}},"verdict_id":"08b03706-e999-40f7-97fc-16d1abfacb79"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:c97500b4c042679afeb0df60827f3b627f5e06b5b804b810c701d0fc1591d906","target":"record","created_at":"2026-05-18T03:10:00Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"a95951fe49943af07040052cbe2b31c810d39c9c6e61cca1684157a13fe4133d","cross_cats_sorted":["cs.CV","cs.LG"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-12T18:09:38Z","title_canon_sha256":"fc124a2f6bfc9f9191983bcdf92128c854775df7e9d78a1b75184569c9a8918c"},"schema_version":"1.0","source":{"id":"2605.12623","kind":"arxiv","version":1}},"canonical_sha256":"646c19dfc06c80b11f405d1b41c62da33bf7c529cf62a53523b492dc102f7538","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"646c19dfc06c80b11f405d1b41c62da33bf7c529cf62a53523b492dc102f7538","first_computed_at":"2026-05-18T03:10:00.376172Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T03:10:00.376172Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"3IRLEK5g+ubeimdwOlKjtdKIujOVqaI0CAl6j/8VxRWJkW3kn5lNaUJvQy8UKrVA8xqelw2Kd66Pxz5fIFvXBw==","signature_status":"signed_v1","signed_at":"2026-05-18T03:10:00.376703Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.12623","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c97500b4c042679afeb0df60827f3b627f5e06b5b804b810c701d0fc1591d906","sha256:6b8b49eb345a41b6f9e054f2bdd21fa1d7fd1af944b7db4e59924ca05e4badd3"],"state_sha256":"adc1ef43117ba96030cec535e6db5ffdc01d3f004143d544906831d1ba2870b8"}