{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2022:XUCPW5FBTH767BFSVERPKJNZMT","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"f9fe0fe6d7cc2103022d4a634ccf54ed67f1aacf40917ca06b4c850164eee961","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-02-16T11:42:34Z","title_canon_sha256":"97fff810efab9ef6eb1704ec0b386efc45d213e0887db251529e3880ec241952"},"schema_version":"1.0","source":{"id":"2202.08005","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2202.08005","created_at":"2026-07-05T05:40:30Z"},{"alias_kind":"arxiv_version","alias_value":"2202.08005v3","created_at":"2026-07-05T05:40:30Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.08005","created_at":"2026-07-05T05:40:30Z"},{"alias_kind":"pith_short_12","alias_value":"XUCPW5FBTH76","created_at":"2026-07-05T05:40:30Z"},{"alias_kind":"pith_short_16","alias_value":"XUCPW5FBTH767BFS","created_at":"2026-07-05T05:40:30Z"},{"alias_kind":"pith_short_8","alias_value":"XUCPW5FB","created_at":"2026-07-05T05:40:30Z"}],"graph_snapshots":[{"event_id":"sha256:5f4a70eb54f3c980f91316c5c45ccf8880aa48d5124c8f413e8f028043a2ef0e","target":"graph","created_at":"2026-07-05T05:40:30Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2202.08005/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Masked language models (MLMs) conventionally mask 15% of tokens due to the belief that more masking would leave insufficient context to learn good representations; this masking rate has been widely used, regardless of model sizes or masking strategies. In this work, we revisit this important choice of MLM pre-training. We first establish that 15% is not universally optimal, and larger models should adopt a higher masking rate. Specifically, we find that masking 40% outperforms 15% for BERT-large size models on GLUE and SQuAD. Interestingly, an extremely high masking rate of 80% can still prese","authors_text":"Alexander Wettig, Danqi Chen, Tianyu Gao, Zexuan Zhong","cross_cats":["cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-02-16T11:42:34Z","title":"Should You Mask 15% in Masked Language Modeling?"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.08005","kind":"arxiv","version":3},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:aa56437fdedca8f7801d7b98dfa0855b62c78b45ddf697c8ebf13dd59f86d03d","target":"record","created_at":"2026-07-05T05:40:30Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"f9fe0fe6d7cc2103022d4a634ccf54ed67f1aacf40917ca06b4c850164eee961","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-02-16T11:42:34Z","title_canon_sha256":"97fff810efab9ef6eb1704ec0b386efc45d213e0887db251529e3880ec241952"},"schema_version":"1.0","source":{"id":"2202.08005","kind":"arxiv","version":3}},"canonical_sha256":"bd04fb74a199ffef84b2a922f525b964d7c006dcfe5f34ce9b3262feae3367d1","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"bd04fb74a199ffef84b2a922f525b964d7c006dcfe5f34ce9b3262feae3367d1","first_computed_at":"2026-07-05T05:40:30.454566Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T05:40:30.454566Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"64UpyD2GRwnxDeIL9lErNJD0cHf2a1dUjNPthqOkLPeEHjy+Etvozua1po8CDPksT1VLmFE8Q92p4CEu6upMDA==","signature_status":"signed_v1","signed_at":"2026-07-05T05:40:30.455051Z","signed_message":"canonical_sha256_bytes"},"source_id":"2202.08005","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:aa56437fdedca8f7801d7b98dfa0855b62c78b45ddf697c8ebf13dd59f86d03d","sha256:5f4a70eb54f3c980f91316c5c45ccf8880aa48d5124c8f413e8f028043a2ef0e"],"state_sha256":"379c985084fe48b30bdbc76decd4f94a632488b6218bf7c73f2c732c979b8ead"}