{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:XTYH5XQ6QT5NRI7HROW6I3ZPIB","short_pith_number":"pith:XTYH5XQ6","schema_version":"1.0","canonical_sha256":"bcf07ede1e84fad8a3e78bade46f2f4055e1e824a6ff49fc31d58d9c8bdb8dd7","source":{"kind":"arxiv","id":"1907.10529","version":3},"attestation_state":"computed","paper":{"title":"SpanBERT: Improving Pre-training by Representing and Predicting Spans","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Daniel S. Weld, Danqi Chen, Luke Zettlemoyer, Mandar Joshi, Omer Levy, Yinhan Liu","submitted_at":"2019-07-24T15:43:40Z","abstract_excerpt":"We present SpanBERT, a pre-training method that is designed to better represent and predict spans of text. Our approach extends BERT by (1) masking contiguous random spans, rather than random tokens, and (2) training the span boundary representations to predict the entire content of the masked span, without relying on the individual token representations within it. SpanBERT consistently outperforms BERT and our better-tuned baselines, with substantial gains on span selection tasks such as question answering and coreference resolution. In particular, with the same training data and model size a"},"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":"1907.10529","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-07-24T15:43:40Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"4d75ee6c126219ddd6ada0ecb0ef65552be36a6b225977e05b13324bf6a76cc7","abstract_canon_sha256":"2484c3d563d6d1a0bbbd1032759473e973d9263f1de09c7475e5776e6d1e3e62"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:34:16.421971Z","signature_b64":"zgFZEXQDRvN0nTEgBDXxdvKUk1hkT9sXbRQdB3ocI5Qlqc5ehaue123hcBXE+WVJX6V4TfsztYgum1DAimwuDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bcf07ede1e84fad8a3e78bade46f2f4055e1e824a6ff49fc31d58d9c8bdb8dd7","last_reissued_at":"2026-07-05T00:34:16.421471Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:34:16.421471Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SpanBERT: Improving Pre-training by Representing and Predicting Spans","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Daniel S. Weld, Danqi Chen, Luke Zettlemoyer, Mandar Joshi, Omer Levy, Yinhan Liu","submitted_at":"2019-07-24T15:43:40Z","abstract_excerpt":"We present SpanBERT, a pre-training method that is designed to better represent and predict spans of text. Our approach extends BERT by (1) masking contiguous random spans, rather than random tokens, and (2) training the span boundary representations to predict the entire content of the masked span, without relying on the individual token representations within it. SpanBERT consistently outperforms BERT and our better-tuned baselines, with substantial gains on span selection tasks such as question answering and coreference resolution. In particular, with the same training data and model size a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1907.10529","kind":"arxiv","version":3},"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/1907.10529/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":"1907.10529","created_at":"2026-07-05T00:34:16.421529+00:00"},{"alias_kind":"arxiv_version","alias_value":"1907.10529v3","created_at":"2026-07-05T00:34:16.421529+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1907.10529","created_at":"2026-07-05T00:34:16.421529+00:00"},{"alias_kind":"pith_short_12","alias_value":"XTYH5XQ6QT5N","created_at":"2026-07-05T00:34:16.421529+00:00"},{"alias_kind":"pith_short_16","alias_value":"XTYH5XQ6QT5NRI7H","created_at":"2026-07-05T00:34:16.421529+00:00"},{"alias_kind":"pith_short_8","alias_value":"XTYH5XQ6","created_at":"2026-07-05T00:34:16.421529+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":9,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.02100","citing_title":"PortBERT: Navigating the Depths of Portuguese Language Models","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2003.10555","citing_title":"ELECTRA: Pre-training Text Encoders as Discriminators Rather Than Generators","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2002.08909","citing_title":"REALM: Retrieval-Augmented Language Model Pre-Training","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"1909.11942","citing_title":"ALBERT: A Lite BERT for Self-supervised Learning of Language Representations","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"1910.13461","citing_title":"BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"1910.10683","citing_title":"Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"1910.03771","citing_title":"HuggingFace's Transformers: State-of-the-art Natural Language Processing","ref_index":158,"is_internal_anchor":false},{"citing_arxiv_id":"1909.08053","citing_title":"Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"1907.11692","citing_title":"RoBERTa: A Robustly Optimized BERT Pretraining Approach","ref_index":19,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XTYH5XQ6QT5NRI7HROW6I3ZPIB","json":"https://pith.science/pith/XTYH5XQ6QT5NRI7HROW6I3ZPIB.json","graph_json":"https://pith.science/api/pith-number/XTYH5XQ6QT5NRI7HROW6I3ZPIB/graph.json","events_json":"https://pith.science/api/pith-number/XTYH5XQ6QT5NRI7HROW6I3ZPIB/events.json","paper":"https://pith.science/paper/XTYH5XQ6"},"agent_actions":{"view_html":"https://pith.science/pith/XTYH5XQ6QT5NRI7HROW6I3ZPIB","download_json":"https://pith.science/pith/XTYH5XQ6QT5NRI7HROW6I3ZPIB.json","view_paper":"https://pith.science/paper/XTYH5XQ6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1907.10529&json=true","fetch_graph":"https://pith.science/api/pith-number/XTYH5XQ6QT5NRI7HROW6I3ZPIB/graph.json","fetch_events":"https://pith.science/api/pith-number/XTYH5XQ6QT5NRI7HROW6I3ZPIB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XTYH5XQ6QT5NRI7HROW6I3ZPIB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XTYH5XQ6QT5NRI7HROW6I3ZPIB/action/storage_attestation","attest_author":"https://pith.science/pith/XTYH5XQ6QT5NRI7HROW6I3ZPIB/action/author_attestation","sign_citation":"https://pith.science/pith/XTYH5XQ6QT5NRI7HROW6I3ZPIB/action/citation_signature","submit_replication":"https://pith.science/pith/XTYH5XQ6QT5NRI7HROW6I3ZPIB/action/replication_record"}},"created_at":"2026-07-05T00:34:16.421529+00:00","updated_at":"2026-07-05T00:34:16.421529+00:00"}