{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:FFM3ZM7T4RF2BLTO32IOREOKIX","short_pith_number":"pith:FFM3ZM7T","schema_version":"1.0","canonical_sha256":"2959bcb3f3e44ba0ae6ede90e891ca45e6967f5a70b90dc648ca18dfaea9156c","source":{"kind":"arxiv","id":"2305.10992","version":1},"attestation_state":"computed","paper":{"title":"How does the task complexity of masked pretraining objectives affect downstream performance?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Atsuki Yamaguchi, Gaku Morio, Hiroaki Ozaki, Terufumi Morishita, Yasuhiro Sogawa","submitted_at":"2023-05-18T14:11:57Z","abstract_excerpt":"Masked language modeling (MLM) is a widely used self-supervised pretraining objective, where a model needs to predict an original token that is replaced with a mask given contexts. Although simpler and computationally efficient pretraining objectives, e.g., predicting the first character of a masked token, have recently shown comparable results to MLM, no objectives with a masking scheme actually outperform it in downstream tasks. Motivated by the assumption that their lack of complexity plays a vital role in the degradation, we validate whether more complex masked objectives can achieve bette"},"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":"2305.10992","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-05-18T14:11:57Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"348c655a0d20c730b5672acacdbeac49ace749a5ec7dec63ae80b8a841cbaad8","abstract_canon_sha256":"de0c59e3459a464755b222eacceb3624a1c25c48772fd5e907d3a3e69216a1dc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:11:29.138185Z","signature_b64":"SOsTaHwPttHv+NP8om1tmcwWtlgvmFj/R1mrHmmWyHdHYQimcSYN+jNAhXVLHpnBn7JDEr/RCcTSPvwqbroCDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2959bcb3f3e44ba0ae6ede90e891ca45e6967f5a70b90dc648ca18dfaea9156c","last_reissued_at":"2026-07-05T06:11:29.137632Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:11:29.137632Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"How does the task complexity of masked pretraining objectives affect downstream performance?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Atsuki Yamaguchi, Gaku Morio, Hiroaki Ozaki, Terufumi Morishita, Yasuhiro Sogawa","submitted_at":"2023-05-18T14:11:57Z","abstract_excerpt":"Masked language modeling (MLM) is a widely used self-supervised pretraining objective, where a model needs to predict an original token that is replaced with a mask given contexts. Although simpler and computationally efficient pretraining objectives, e.g., predicting the first character of a masked token, have recently shown comparable results to MLM, no objectives with a masking scheme actually outperform it in downstream tasks. Motivated by the assumption that their lack of complexity plays a vital role in the degradation, we validate whether more complex masked objectives can achieve bette"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.10992","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/2305.10992/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":"2305.10992","created_at":"2026-07-05T06:11:29.137692+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.10992v1","created_at":"2026-07-05T06:11:29.137692+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.10992","created_at":"2026-07-05T06:11:29.137692+00:00"},{"alias_kind":"pith_short_12","alias_value":"FFM3ZM7T4RF2","created_at":"2026-07-05T06:11:29.137692+00:00"},{"alias_kind":"pith_short_16","alias_value":"FFM3ZM7T4RF2BLTO","created_at":"2026-07-05T06:11:29.137692+00:00"},{"alias_kind":"pith_short_8","alias_value":"FFM3ZM7T","created_at":"2026-07-05T06:11:29.137692+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2608.03855","citing_title":"Bi-semantic Chemical Embedder for Joint Representation Learning of SMILES and Natural Language","ref_index":33,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FFM3ZM7T4RF2BLTO32IOREOKIX","json":"https://pith.science/pith/FFM3ZM7T4RF2BLTO32IOREOKIX.json","graph_json":"https://pith.science/api/pith-number/FFM3ZM7T4RF2BLTO32IOREOKIX/graph.json","events_json":"https://pith.science/api/pith-number/FFM3ZM7T4RF2BLTO32IOREOKIX/events.json","paper":"https://pith.science/paper/FFM3ZM7T"},"agent_actions":{"view_html":"https://pith.science/pith/FFM3ZM7T4RF2BLTO32IOREOKIX","download_json":"https://pith.science/pith/FFM3ZM7T4RF2BLTO32IOREOKIX.json","view_paper":"https://pith.science/paper/FFM3ZM7T","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.10992&json=true","fetch_graph":"https://pith.science/api/pith-number/FFM3ZM7T4RF2BLTO32IOREOKIX/graph.json","fetch_events":"https://pith.science/api/pith-number/FFM3ZM7T4RF2BLTO32IOREOKIX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FFM3ZM7T4RF2BLTO32IOREOKIX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FFM3ZM7T4RF2BLTO32IOREOKIX/action/storage_attestation","attest_author":"https://pith.science/pith/FFM3ZM7T4RF2BLTO32IOREOKIX/action/author_attestation","sign_citation":"https://pith.science/pith/FFM3ZM7T4RF2BLTO32IOREOKIX/action/citation_signature","submit_replication":"https://pith.science/pith/FFM3ZM7T4RF2BLTO32IOREOKIX/action/replication_record"}},"created_at":"2026-07-05T06:11:29.137692+00:00","updated_at":"2026-07-05T06:11:29.137692+00:00"}