{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:I757PMDYBXMPBL5NUPA47TMDGW","short_pith_number":"pith:I757PMDY","schema_version":"1.0","canonical_sha256":"47fbf7b0780dd8f0afada3c1cfcd8335abb9ba7ba4414262b6e4cb82b909ba31","source":{"kind":"arxiv","id":"2106.06381","version":2},"attestation_state":"computed","paper":{"title":"Improving Pretrained Cross-Lingual Language Models via Self-Labeled Word Alignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bo Zheng, Furu Wei, Heyan Huang, Li Dong, Shaohan Huang, Xian-Ling Mao, Zewen Chi","submitted_at":"2021-06-11T13:36:01Z","abstract_excerpt":"The cross-lingual language models are typically pretrained with masked language modeling on multilingual text or parallel sentences. In this paper, we introduce denoising word alignment as a new cross-lingual pre-training task. Specifically, the model first self-labels word alignments for parallel sentences. Then we randomly mask tokens in a bitext pair. Given a masked token, the model uses a pointer network to predict the aligned token in the other language. We alternately perform the above two steps in an expectation-maximization manner. Experimental results show that our method improves cro"},"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":"2106.06381","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-06-11T13:36:01Z","cross_cats_sorted":[],"title_canon_sha256":"468707409e86c2569e9ffb4307edcf8327c29a4fc264afc0d0ed24ae28a1f784","abstract_canon_sha256":"33f8c986e5e31c8f7525ecfd25ca5f20d9be0f329feb1ff6a39e62a43dddc3e3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:13:29.824615Z","signature_b64":"vrQOpq9+3G8W+aGdC73qrpNlNLVkuS/S2EUHaFuVkWDfLYhW1FqfgsZx2Q8TTEbUyO8SjvURrQPKmMim4/ogDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"47fbf7b0780dd8f0afada3c1cfcd8335abb9ba7ba4414262b6e4cb82b909ba31","last_reissued_at":"2026-07-05T03:13:29.824141Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:13:29.824141Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Improving Pretrained Cross-Lingual Language Models via Self-Labeled Word Alignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bo Zheng, Furu Wei, Heyan Huang, Li Dong, Shaohan Huang, Xian-Ling Mao, Zewen Chi","submitted_at":"2021-06-11T13:36:01Z","abstract_excerpt":"The cross-lingual language models are typically pretrained with masked language modeling on multilingual text or parallel sentences. In this paper, we introduce denoising word alignment as a new cross-lingual pre-training task. Specifically, the model first self-labels word alignments for parallel sentences. Then we randomly mask tokens in a bitext pair. Given a masked token, the model uses a pointer network to predict the aligned token in the other language. We alternately perform the above two steps in an expectation-maximization manner. Experimental results show that our method improves cro"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2106.06381","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/2106.06381/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":"2106.06381","created_at":"2026-07-05T03:13:29.824198+00:00"},{"alias_kind":"arxiv_version","alias_value":"2106.06381v2","created_at":"2026-07-05T03:13:29.824198+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2106.06381","created_at":"2026-07-05T03:13:29.824198+00:00"},{"alias_kind":"pith_short_12","alias_value":"I757PMDYBXMP","created_at":"2026-07-05T03:13:29.824198+00:00"},{"alias_kind":"pith_short_16","alias_value":"I757PMDYBXMPBL5N","created_at":"2026-07-05T03:13:29.824198+00:00"},{"alias_kind":"pith_short_8","alias_value":"I757PMDY","created_at":"2026-07-05T03:13:29.824198+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/I757PMDYBXMPBL5NUPA47TMDGW","json":"https://pith.science/pith/I757PMDYBXMPBL5NUPA47TMDGW.json","graph_json":"https://pith.science/api/pith-number/I757PMDYBXMPBL5NUPA47TMDGW/graph.json","events_json":"https://pith.science/api/pith-number/I757PMDYBXMPBL5NUPA47TMDGW/events.json","paper":"https://pith.science/paper/I757PMDY"},"agent_actions":{"view_html":"https://pith.science/pith/I757PMDYBXMPBL5NUPA47TMDGW","download_json":"https://pith.science/pith/I757PMDYBXMPBL5NUPA47TMDGW.json","view_paper":"https://pith.science/paper/I757PMDY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2106.06381&json=true","fetch_graph":"https://pith.science/api/pith-number/I757PMDYBXMPBL5NUPA47TMDGW/graph.json","fetch_events":"https://pith.science/api/pith-number/I757PMDYBXMPBL5NUPA47TMDGW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/I757PMDYBXMPBL5NUPA47TMDGW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/I757PMDYBXMPBL5NUPA47TMDGW/action/storage_attestation","attest_author":"https://pith.science/pith/I757PMDYBXMPBL5NUPA47TMDGW/action/author_attestation","sign_citation":"https://pith.science/pith/I757PMDYBXMPBL5NUPA47TMDGW/action/citation_signature","submit_replication":"https://pith.science/pith/I757PMDYBXMPBL5NUPA47TMDGW/action/replication_record"}},"created_at":"2026-07-05T03:13:29.824198+00:00","updated_at":"2026-07-05T03:13:29.824198+00:00"}