{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:D7DPK2NNP75IFZ2YVM2EMBER5Y","short_pith_number":"pith:D7DPK2NN","schema_version":"1.0","canonical_sha256":"1fc6f569ad7ffa82e758ab34460491ee1d34b51dc5971d49cba3a26e77752a96","source":{"kind":"arxiv","id":"1911.02972","version":2},"attestation_state":"computed","paper":{"title":"Blockwise Self-Attention for Long Document Understanding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Hao Ma, Jie Tang, Jiezhong Qiu, Omer Levy, Scott Wen-tau Yih, Sinong Wang","submitted_at":"2019-11-07T16:35:53Z","abstract_excerpt":"We present BlockBERT, a lightweight and efficient BERT model for better modeling long-distance dependencies. Our model extends BERT by introducing sparse block structures into the attention matrix to reduce both memory consumption and training/inference time, which also enables attention heads to capture either short- or long-range contextual information. We conduct experiments on language model pre-training and several benchmark question answering datasets with various paragraph lengths. BlockBERT uses 18.7-36.1% less memory and 12.0-25.1% less time to learn the model. During testing, BlockBE"},"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":"1911.02972","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-11-07T16:35:53Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"ecbe03c3e25b7c3ee78ee4078fc4ce4a30efc2d3f733955c2d36b87b775523bd","abstract_canon_sha256":"9af55e94dc6d9d8fabbe41bd90a9250bcf017d7ffd3edcf77ef2f85fa3ed0929"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:48:07.048555Z","signature_b64":"IcCwghXSjuypxmfNp/F+NfCvoyX90hzvnagBUh1dW+nAlicmzsVBiG1M7r3BFkzhRIcTGMGn0EoEp2rOhWujCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1fc6f569ad7ffa82e758ab34460491ee1d34b51dc5971d49cba3a26e77752a96","last_reissued_at":"2026-07-05T01:48:07.048048Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:48:07.048048Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Blockwise Self-Attention for Long Document Understanding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Hao Ma, Jie Tang, Jiezhong Qiu, Omer Levy, Scott Wen-tau Yih, Sinong Wang","submitted_at":"2019-11-07T16:35:53Z","abstract_excerpt":"We present BlockBERT, a lightweight and efficient BERT model for better modeling long-distance dependencies. Our model extends BERT by introducing sparse block structures into the attention matrix to reduce both memory consumption and training/inference time, which also enables attention heads to capture either short- or long-range contextual information. We conduct experiments on language model pre-training and several benchmark question answering datasets with various paragraph lengths. BlockBERT uses 18.7-36.1% less memory and 12.0-25.1% less time to learn the model. During testing, BlockBE"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1911.02972","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/1911.02972/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":"1911.02972","created_at":"2026-07-05T01:48:07.048104+00:00"},{"alias_kind":"arxiv_version","alias_value":"1911.02972v2","created_at":"2026-07-05T01:48:07.048104+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1911.02972","created_at":"2026-07-05T01:48:07.048104+00:00"},{"alias_kind":"pith_short_12","alias_value":"D7DPK2NNP75I","created_at":"2026-07-05T01:48:07.048104+00:00"},{"alias_kind":"pith_short_16","alias_value":"D7DPK2NNP75IFZ2Y","created_at":"2026-07-05T01:48:07.048104+00:00"},{"alias_kind":"pith_short_8","alias_value":"D7DPK2NN","created_at":"2026-07-05T01:48:07.048104+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2502.13189","citing_title":"MoBA: Mixture of Block Attention for Long-Context LLMs","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2006.04768","citing_title":"Linformer: Self-Attention with Linear Complexity","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2010.04159","citing_title":"Deformable DETR: Deformable Transformers for End-to-End Object Detection","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2004.05150","citing_title":"Longformer: The Long-Document Transformer","ref_index":111,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/D7DPK2NNP75IFZ2YVM2EMBER5Y","json":"https://pith.science/pith/D7DPK2NNP75IFZ2YVM2EMBER5Y.json","graph_json":"https://pith.science/api/pith-number/D7DPK2NNP75IFZ2YVM2EMBER5Y/graph.json","events_json":"https://pith.science/api/pith-number/D7DPK2NNP75IFZ2YVM2EMBER5Y/events.json","paper":"https://pith.science/paper/D7DPK2NN"},"agent_actions":{"view_html":"https://pith.science/pith/D7DPK2NNP75IFZ2YVM2EMBER5Y","download_json":"https://pith.science/pith/D7DPK2NNP75IFZ2YVM2EMBER5Y.json","view_paper":"https://pith.science/paper/D7DPK2NN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1911.02972&json=true","fetch_graph":"https://pith.science/api/pith-number/D7DPK2NNP75IFZ2YVM2EMBER5Y/graph.json","fetch_events":"https://pith.science/api/pith-number/D7DPK2NNP75IFZ2YVM2EMBER5Y/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/D7DPK2NNP75IFZ2YVM2EMBER5Y/action/timestamp_anchor","attest_storage":"https://pith.science/pith/D7DPK2NNP75IFZ2YVM2EMBER5Y/action/storage_attestation","attest_author":"https://pith.science/pith/D7DPK2NNP75IFZ2YVM2EMBER5Y/action/author_attestation","sign_citation":"https://pith.science/pith/D7DPK2NNP75IFZ2YVM2EMBER5Y/action/citation_signature","submit_replication":"https://pith.science/pith/D7DPK2NNP75IFZ2YVM2EMBER5Y/action/replication_record"}},"created_at":"2026-07-05T01:48:07.048104+00:00","updated_at":"2026-07-05T01:48:07.048104+00:00"}