{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:PTI4SJLKECOYZFVGBLZLDC4O4F","short_pith_number":"pith:PTI4SJLK","schema_version":"1.0","canonical_sha256":"7cd1c9256a209d8c96a60af2b18b8ee14e3fb7cdd39bc151471161dd45fa4a8b","source":{"kind":"arxiv","id":"2002.12804","version":1},"attestation_state":"computed","paper":{"title":"UniLMv2: Pseudo-Masked Language Models for Unified Language Model Pre-Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Furu Wei, Hangbo Bao, Hsiao-Wuen Hon, Jianfeng Gao, Li Dong, Ming Zhou, Nan Yang, Songhao Piao, Wenhui Wang, XiaoDong Liu, Yu Wang","submitted_at":"2020-02-28T15:28:49Z","abstract_excerpt":"We propose to pre-train a unified language model for both autoencoding and partially autoregressive language modeling tasks using a novel training procedure, referred to as a pseudo-masked language model (PMLM). Given an input text with masked tokens, we rely on conventional masks to learn inter-relations between corrupted tokens and context via autoencoding, and pseudo masks to learn intra-relations between masked spans via partially autoregressive modeling. With well-designed position embeddings and self-attention masks, the context encodings are reused to avoid redundant computation. Moreov"},"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":"2002.12804","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-02-28T15:28:49Z","cross_cats_sorted":[],"title_canon_sha256":"ac6b139ccd36044f31f0ad47052f1a34c3273126c9e0c1ac61bff3dac3282587","abstract_canon_sha256":"c32700ba127e7734346c830105e78ee0ec7203845707d37a34b86d955cccbea1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:44:34.423953Z","signature_b64":"RDCqrxkzmw36v8LeQ2p2Rshb8qa/vupVUxzNqpFATd0KTXAypl/LAcZki8V63R/4zQEMjM2NZoJFXxUFZjoGDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7cd1c9256a209d8c96a60af2b18b8ee14e3fb7cdd39bc151471161dd45fa4a8b","last_reissued_at":"2026-07-05T00:44:34.423578Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:44:34.423578Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"UniLMv2: Pseudo-Masked Language Models for Unified Language Model Pre-Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Furu Wei, Hangbo Bao, Hsiao-Wuen Hon, Jianfeng Gao, Li Dong, Ming Zhou, Nan Yang, Songhao Piao, Wenhui Wang, XiaoDong Liu, Yu Wang","submitted_at":"2020-02-28T15:28:49Z","abstract_excerpt":"We propose to pre-train a unified language model for both autoencoding and partially autoregressive language modeling tasks using a novel training procedure, referred to as a pseudo-masked language model (PMLM). Given an input text with masked tokens, we rely on conventional masks to learn inter-relations between corrupted tokens and context via autoencoding, and pseudo masks to learn intra-relations between masked spans via partially autoregressive modeling. With well-designed position embeddings and self-attention masks, the context encodings are reused to avoid redundant computation. Moreov"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2002.12804","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/2002.12804/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":"2002.12804","created_at":"2026-07-05T00:44:34.423635+00:00"},{"alias_kind":"arxiv_version","alias_value":"2002.12804v1","created_at":"2026-07-05T00:44:34.423635+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2002.12804","created_at":"2026-07-05T00:44:34.423635+00:00"},{"alias_kind":"pith_short_12","alias_value":"PTI4SJLKECOY","created_at":"2026-07-05T00:44:34.423635+00:00"},{"alias_kind":"pith_short_16","alias_value":"PTI4SJLKECOYZFVG","created_at":"2026-07-05T00:44:34.423635+00:00"},{"alias_kind":"pith_short_8","alias_value":"PTI4SJLK","created_at":"2026-07-05T00:44:34.423635+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.13535","citing_title":"Information Extraction from Visually Rich Documents using LLM-based Organization of Documents into Independent Textual Segments","ref_index":2020,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PTI4SJLKECOYZFVGBLZLDC4O4F","json":"https://pith.science/pith/PTI4SJLKECOYZFVGBLZLDC4O4F.json","graph_json":"https://pith.science/api/pith-number/PTI4SJLKECOYZFVGBLZLDC4O4F/graph.json","events_json":"https://pith.science/api/pith-number/PTI4SJLKECOYZFVGBLZLDC4O4F/events.json","paper":"https://pith.science/paper/PTI4SJLK"},"agent_actions":{"view_html":"https://pith.science/pith/PTI4SJLKECOYZFVGBLZLDC4O4F","download_json":"https://pith.science/pith/PTI4SJLKECOYZFVGBLZLDC4O4F.json","view_paper":"https://pith.science/paper/PTI4SJLK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2002.12804&json=true","fetch_graph":"https://pith.science/api/pith-number/PTI4SJLKECOYZFVGBLZLDC4O4F/graph.json","fetch_events":"https://pith.science/api/pith-number/PTI4SJLKECOYZFVGBLZLDC4O4F/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PTI4SJLKECOYZFVGBLZLDC4O4F/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PTI4SJLKECOYZFVGBLZLDC4O4F/action/storage_attestation","attest_author":"https://pith.science/pith/PTI4SJLKECOYZFVGBLZLDC4O4F/action/author_attestation","sign_citation":"https://pith.science/pith/PTI4SJLKECOYZFVGBLZLDC4O4F/action/citation_signature","submit_replication":"https://pith.science/pith/PTI4SJLKECOYZFVGBLZLDC4O4F/action/replication_record"}},"created_at":"2026-07-05T00:44:34.423635+00:00","updated_at":"2026-07-05T00:44:34.423635+00:00"}