{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:TZQMRINH7NJZLHHAARLEYFQAFM","short_pith_number":"pith:TZQMRINH","schema_version":"1.0","canonical_sha256":"9e60c8a1a7fb53959ce004564c16002b09b58f11d13155555ac796b201e17770","source":{"kind":"arxiv","id":"2507.06761","version":1},"attestation_state":"computed","paper":{"title":"Finetuning Vision-Language Models as OCR Systems for Low-Resource Languages: A Case Study of Manchu","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Donghyeok Choi, Yan Hon Michael Chung","submitted_at":"2025-07-09T11:38:20Z","abstract_excerpt":"Manchu, a critically endangered language essential for understanding early modern Eastern Eurasian history, lacks effective OCR systems that can handle real-world historical documents. This study develops high-performing OCR systems by fine-tuning three open-source vision-language models (LLaMA-3.2-11B, Qwen2.5-VL-7B, Qwen2.5-VL-3B) on 60,000 synthetic Manchu word images using parameter-efficient training. LLaMA-3.2-11B achieved exceptional performance with 98.3\\% word accuracy and 0.0024 character error rate on synthetic data, while crucially maintaining 93.1\\% accuracy on real-world handwrit"},"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":"2507.06761","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-07-09T11:38:20Z","cross_cats_sorted":[],"title_canon_sha256":"e366c40ea4baefe8a563415157afbc1c3e502671786ba62f363e3eeafd4b9a4e","abstract_canon_sha256":"be079054c7912d79740c7eedb13c589b5af0f37bc831c215fb95b3c555fb7f5a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:34:22.778920Z","signature_b64":"jeO5Qvzv/NJS0b2Qy3eBlGedLRUL0PwsGTGqKi61o5R8TEOA6UY7R7r06JNJgcdbjIORq9wOnmFQFilzPc3NAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9e60c8a1a7fb53959ce004564c16002b09b58f11d13155555ac796b201e17770","last_reissued_at":"2026-07-05T11:34:22.778424Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:34:22.778424Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Finetuning Vision-Language Models as OCR Systems for Low-Resource Languages: A Case Study of Manchu","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Donghyeok Choi, Yan Hon Michael Chung","submitted_at":"2025-07-09T11:38:20Z","abstract_excerpt":"Manchu, a critically endangered language essential for understanding early modern Eastern Eurasian history, lacks effective OCR systems that can handle real-world historical documents. This study develops high-performing OCR systems by fine-tuning three open-source vision-language models (LLaMA-3.2-11B, Qwen2.5-VL-7B, Qwen2.5-VL-3B) on 60,000 synthetic Manchu word images using parameter-efficient training. LLaMA-3.2-11B achieved exceptional performance with 98.3\\% word accuracy and 0.0024 character error rate on synthetic data, while crucially maintaining 93.1\\% accuracy on real-world handwrit"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.06761","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/2507.06761/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":"2507.06761","created_at":"2026-07-05T11:34:22.778490+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.06761v1","created_at":"2026-07-05T11:34:22.778490+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.06761","created_at":"2026-07-05T11:34:22.778490+00:00"},{"alias_kind":"pith_short_12","alias_value":"TZQMRINH7NJZ","created_at":"2026-07-05T11:34:22.778490+00:00"},{"alias_kind":"pith_short_16","alias_value":"TZQMRINH7NJZLHHA","created_at":"2026-07-05T11:34:22.778490+00:00"},{"alias_kind":"pith_short_8","alias_value":"TZQMRINH","created_at":"2026-07-05T11:34:22.778490+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.12623","citing_title":"DocAtlas: Multilingual Document Understanding Across 80+ Languages","ref_index":64,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12623","citing_title":"DocAtlas: Multilingual Document Understanding Across 80+ Languages","ref_index":71,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TZQMRINH7NJZLHHAARLEYFQAFM","json":"https://pith.science/pith/TZQMRINH7NJZLHHAARLEYFQAFM.json","graph_json":"https://pith.science/api/pith-number/TZQMRINH7NJZLHHAARLEYFQAFM/graph.json","events_json":"https://pith.science/api/pith-number/TZQMRINH7NJZLHHAARLEYFQAFM/events.json","paper":"https://pith.science/paper/TZQMRINH"},"agent_actions":{"view_html":"https://pith.science/pith/TZQMRINH7NJZLHHAARLEYFQAFM","download_json":"https://pith.science/pith/TZQMRINH7NJZLHHAARLEYFQAFM.json","view_paper":"https://pith.science/paper/TZQMRINH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.06761&json=true","fetch_graph":"https://pith.science/api/pith-number/TZQMRINH7NJZLHHAARLEYFQAFM/graph.json","fetch_events":"https://pith.science/api/pith-number/TZQMRINH7NJZLHHAARLEYFQAFM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TZQMRINH7NJZLHHAARLEYFQAFM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TZQMRINH7NJZLHHAARLEYFQAFM/action/storage_attestation","attest_author":"https://pith.science/pith/TZQMRINH7NJZLHHAARLEYFQAFM/action/author_attestation","sign_citation":"https://pith.science/pith/TZQMRINH7NJZLHHAARLEYFQAFM/action/citation_signature","submit_replication":"https://pith.science/pith/TZQMRINH7NJZLHHAARLEYFQAFM/action/replication_record"}},"created_at":"2026-07-05T11:34:22.778490+00:00","updated_at":"2026-07-05T11:34:22.778490+00:00"}