{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:B563IOP3M5NLLXVVPCMLLOCBP5","short_pith_number":"pith:B563IOP3","schema_version":"1.0","canonical_sha256":"0f7db439fb675ab5deb57898b5b8417f4eecfe77b9defb588ef55c7c56d0b989","source":{"kind":"arxiv","id":"2109.07020","version":1},"attestation_state":"computed","paper":{"title":"Frequency Effects on Syntactic Rule Learning in Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dan Garrette, Ellie Pavlick, Jason Wei, Tal Linzen","submitted_at":"2021-09-14T23:45:20Z","abstract_excerpt":"Pre-trained language models perform well on a variety of linguistic tasks that require symbolic reasoning, raising the question of whether such models implicitly represent abstract symbols and rules. We investigate this question using the case study of BERT's performance on English subject-verb agreement. Unlike prior work, we train multiple instances of BERT from scratch, allowing us to perform a series of controlled interventions at pre-training time. We show that BERT often generalizes well to subject-verb pairs that never occurred in training, suggesting a degree of rule-governed behavior."},"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":"2109.07020","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-09-14T23:45:20Z","cross_cats_sorted":[],"title_canon_sha256":"5cafddfad038338c7f6f0696fe67dd44454d28b3dd9490cafe760e37d945da4a","abstract_canon_sha256":"b24b92da9c47f565470b4f360aacf6fe1aab48711d63ebc9bb45c78b0c948df4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:14:38.119449Z","signature_b64":"JWBubVT1dvlNWAtGhhnvWwqPhdU3skMLyBk2qmVD5pEX/FvRYXQuILsB1FdSPxZeF33QFYnbf8Jo+5SV1lzrDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0f7db439fb675ab5deb57898b5b8417f4eecfe77b9defb588ef55c7c56d0b989","last_reissued_at":"2026-07-05T03:14:38.119025Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:14:38.119025Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Frequency Effects on Syntactic Rule Learning in Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dan Garrette, Ellie Pavlick, Jason Wei, Tal Linzen","submitted_at":"2021-09-14T23:45:20Z","abstract_excerpt":"Pre-trained language models perform well on a variety of linguistic tasks that require symbolic reasoning, raising the question of whether such models implicitly represent abstract symbols and rules. We investigate this question using the case study of BERT's performance on English subject-verb agreement. Unlike prior work, we train multiple instances of BERT from scratch, allowing us to perform a series of controlled interventions at pre-training time. We show that BERT often generalizes well to subject-verb pairs that never occurred in training, suggesting a degree of rule-governed behavior."},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2109.07020","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/2109.07020/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":"2109.07020","created_at":"2026-07-05T03:14:38.119082+00:00"},{"alias_kind":"arxiv_version","alias_value":"2109.07020v1","created_at":"2026-07-05T03:14:38.119082+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2109.07020","created_at":"2026-07-05T03:14:38.119082+00:00"},{"alias_kind":"pith_short_12","alias_value":"B563IOP3M5NL","created_at":"2026-07-05T03:14:38.119082+00:00"},{"alias_kind":"pith_short_16","alias_value":"B563IOP3M5NLLXVV","created_at":"2026-07-05T03:14:38.119082+00:00"},{"alias_kind":"pith_short_8","alias_value":"B563IOP3","created_at":"2026-07-05T03:14:38.119082+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.27460","citing_title":"Developmental approach reveals the statistical learning of Neural Language Models: Transformers generalize from the most abstract statistical patterns","ref_index":58,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/B563IOP3M5NLLXVVPCMLLOCBP5","json":"https://pith.science/pith/B563IOP3M5NLLXVVPCMLLOCBP5.json","graph_json":"https://pith.science/api/pith-number/B563IOP3M5NLLXVVPCMLLOCBP5/graph.json","events_json":"https://pith.science/api/pith-number/B563IOP3M5NLLXVVPCMLLOCBP5/events.json","paper":"https://pith.science/paper/B563IOP3"},"agent_actions":{"view_html":"https://pith.science/pith/B563IOP3M5NLLXVVPCMLLOCBP5","download_json":"https://pith.science/pith/B563IOP3M5NLLXVVPCMLLOCBP5.json","view_paper":"https://pith.science/paper/B563IOP3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2109.07020&json=true","fetch_graph":"https://pith.science/api/pith-number/B563IOP3M5NLLXVVPCMLLOCBP5/graph.json","fetch_events":"https://pith.science/api/pith-number/B563IOP3M5NLLXVVPCMLLOCBP5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/B563IOP3M5NLLXVVPCMLLOCBP5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/B563IOP3M5NLLXVVPCMLLOCBP5/action/storage_attestation","attest_author":"https://pith.science/pith/B563IOP3M5NLLXVVPCMLLOCBP5/action/author_attestation","sign_citation":"https://pith.science/pith/B563IOP3M5NLLXVVPCMLLOCBP5/action/citation_signature","submit_replication":"https://pith.science/pith/B563IOP3M5NLLXVVPCMLLOCBP5/action/replication_record"}},"created_at":"2026-07-05T03:14:38.119082+00:00","updated_at":"2026-07-05T03:14:38.119082+00:00"}