{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:FBG5LJIRGMSBMMLGGN46BWADHD","short_pith_number":"pith:FBG5LJIR","schema_version":"1.0","canonical_sha256":"284dd5a51133241631663379e0d80338d1c5d77500ad69f38475eeb4cc592fff","source":{"kind":"arxiv","id":"2301.04761","version":2},"attestation_state":"computed","paper":{"title":"NarrowBERT: Accelerating Masked Language Model Pretraining and Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Daniel Cheng, Haoxin Li, Jungo Kasai, Noah A. Smith, Phillip Keung","submitted_at":"2023-01-11T23:45:50Z","abstract_excerpt":"Large-scale language model pretraining is a very successful form of self-supervised learning in natural language processing, but it is increasingly expensive to perform as the models and pretraining corpora have become larger over time. We propose NarrowBERT, a modified transformer encoder that increases the throughput for masked language model pretraining by more than $2\\times$. NarrowBERT sparsifies the transformer model such that the self-attention queries and feedforward layers only operate on the masked tokens of each sentence during pretraining, rather than all of the tokens as with the "},"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":"2301.04761","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-01-11T23:45:50Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"c85bb2315e9650a6004e9cd2d97dbb3e1e964aeb2079aa82408d5bf18867e4dc","abstract_canon_sha256":"97a962a59260cb30fc88d25cebf2ad63b0aebdc267b9a79c872bc3b290a720ea"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:17:44.607223Z","signature_b64":"0V7NdGMuyyiKGBVnZVZ2cjChT/cRhNTMMsOO9VqYxx95R1v9WkXh/fNM6qUT+8Z6hXLudTGcKc+pvx329Z9kCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"284dd5a51133241631663379e0d80338d1c5d77500ad69f38475eeb4cc592fff","last_reissued_at":"2026-07-05T06:17:44.606742Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:17:44.606742Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"NarrowBERT: Accelerating Masked Language Model Pretraining and Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Daniel Cheng, Haoxin Li, Jungo Kasai, Noah A. Smith, Phillip Keung","submitted_at":"2023-01-11T23:45:50Z","abstract_excerpt":"Large-scale language model pretraining is a very successful form of self-supervised learning in natural language processing, but it is increasingly expensive to perform as the models and pretraining corpora have become larger over time. We propose NarrowBERT, a modified transformer encoder that increases the throughput for masked language model pretraining by more than $2\\times$. NarrowBERT sparsifies the transformer model such that the self-attention queries and feedforward layers only operate on the masked tokens of each sentence during pretraining, rather than all of the tokens as with the "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2301.04761","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/2301.04761/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":"2301.04761","created_at":"2026-07-05T06:17:44.606813+00:00"},{"alias_kind":"arxiv_version","alias_value":"2301.04761v2","created_at":"2026-07-05T06:17:44.606813+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2301.04761","created_at":"2026-07-05T06:17:44.606813+00:00"},{"alias_kind":"pith_short_12","alias_value":"FBG5LJIRGMSB","created_at":"2026-07-05T06:17:44.606813+00:00"},{"alias_kind":"pith_short_16","alias_value":"FBG5LJIRGMSBMMLG","created_at":"2026-07-05T06:17:44.606813+00:00"},{"alias_kind":"pith_short_8","alias_value":"FBG5LJIR","created_at":"2026-07-05T06:17:44.606813+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/FBG5LJIRGMSBMMLGGN46BWADHD","json":"https://pith.science/pith/FBG5LJIRGMSBMMLGGN46BWADHD.json","graph_json":"https://pith.science/api/pith-number/FBG5LJIRGMSBMMLGGN46BWADHD/graph.json","events_json":"https://pith.science/api/pith-number/FBG5LJIRGMSBMMLGGN46BWADHD/events.json","paper":"https://pith.science/paper/FBG5LJIR"},"agent_actions":{"view_html":"https://pith.science/pith/FBG5LJIRGMSBMMLGGN46BWADHD","download_json":"https://pith.science/pith/FBG5LJIRGMSBMMLGGN46BWADHD.json","view_paper":"https://pith.science/paper/FBG5LJIR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2301.04761&json=true","fetch_graph":"https://pith.science/api/pith-number/FBG5LJIRGMSBMMLGGN46BWADHD/graph.json","fetch_events":"https://pith.science/api/pith-number/FBG5LJIRGMSBMMLGGN46BWADHD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FBG5LJIRGMSBMMLGGN46BWADHD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FBG5LJIRGMSBMMLGGN46BWADHD/action/storage_attestation","attest_author":"https://pith.science/pith/FBG5LJIRGMSBMMLGGN46BWADHD/action/author_attestation","sign_citation":"https://pith.science/pith/FBG5LJIRGMSBMMLGGN46BWADHD/action/citation_signature","submit_replication":"https://pith.science/pith/FBG5LJIRGMSBMMLGGN46BWADHD/action/replication_record"}},"created_at":"2026-07-05T06:17:44.606813+00:00","updated_at":"2026-07-05T06:17:44.606813+00:00"}