{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:OXS46RLFZNRL3CTHKYB6O43EA4","short_pith_number":"pith:OXS46RLF","schema_version":"1.0","canonical_sha256":"75e5cf4565cb62bd8a675603e77364070b774ecd81cf2f1a8c82a4b2b632fb45","source":{"kind":"arxiv","id":"1907.13528","version":2},"attestation_state":"computed","paper":{"title":"What BERT is not: Lessons from a new suite of psycholinguistic diagnostics for language models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Allyson Ettinger","submitted_at":"2019-07-31T14:37:32Z","abstract_excerpt":"Pre-training by language modeling has become a popular and successful approach to NLP tasks, but we have yet to understand exactly what linguistic capacities these pre-training processes confer upon models. In this paper we introduce a suite of diagnostics drawn from human language experiments, which allow us to ask targeted questions about the information used by language models for generating predictions in context. As a case study, we apply these diagnostics to the popular BERT model, finding that it can generally distinguish good from bad completions involving shared category or role rever"},"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":"1907.13528","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-07-31T14:37:32Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"a74629845895a300154d836258cadb5f3b04ebfbcdf8323a119d9564d9a99ad7","abstract_canon_sha256":"7798bdfc2662ec3b7dbf2be90f11a90c67492c6fa86e1edf57bde88f0f8f17e2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:17:57.396463Z","signature_b64":"GGbc81evHIB312+bvjDMNeNsYoOksZpd5QjhIjWmMVHitQRYn0WRLcm7UmmmLdcttXhid3E2a/ySjERgacCBAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"75e5cf4565cb62bd8a675603e77364070b774ecd81cf2f1a8c82a4b2b632fb45","last_reissued_at":"2026-07-05T01:17:57.396011Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:17:57.396011Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"What BERT is not: Lessons from a new suite of psycholinguistic diagnostics for language models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Allyson Ettinger","submitted_at":"2019-07-31T14:37:32Z","abstract_excerpt":"Pre-training by language modeling has become a popular and successful approach to NLP tasks, but we have yet to understand exactly what linguistic capacities these pre-training processes confer upon models. In this paper we introduce a suite of diagnostics drawn from human language experiments, which allow us to ask targeted questions about the information used by language models for generating predictions in context. As a case study, we apply these diagnostics to the popular BERT model, finding that it can generally distinguish good from bad completions involving shared category or role rever"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1907.13528","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/1907.13528/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":"1907.13528","created_at":"2026-07-05T01:17:57.396071+00:00"},{"alias_kind":"arxiv_version","alias_value":"1907.13528v2","created_at":"2026-07-05T01:17:57.396071+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1907.13528","created_at":"2026-07-05T01:17:57.396071+00:00"},{"alias_kind":"pith_short_12","alias_value":"OXS46RLFZNRL","created_at":"2026-07-05T01:17:57.396071+00:00"},{"alias_kind":"pith_short_16","alias_value":"OXS46RLFZNRL3CTH","created_at":"2026-07-05T01:17:57.396071+00:00"},{"alias_kind":"pith_short_8","alias_value":"OXS46RLF","created_at":"2026-07-05T01:17:57.396071+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.02100","citing_title":"PortBERT: Navigating the Depths of Portuguese Language Models","ref_index":16,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OXS46RLFZNRL3CTHKYB6O43EA4","json":"https://pith.science/pith/OXS46RLFZNRL3CTHKYB6O43EA4.json","graph_json":"https://pith.science/api/pith-number/OXS46RLFZNRL3CTHKYB6O43EA4/graph.json","events_json":"https://pith.science/api/pith-number/OXS46RLFZNRL3CTHKYB6O43EA4/events.json","paper":"https://pith.science/paper/OXS46RLF"},"agent_actions":{"view_html":"https://pith.science/pith/OXS46RLFZNRL3CTHKYB6O43EA4","download_json":"https://pith.science/pith/OXS46RLFZNRL3CTHKYB6O43EA4.json","view_paper":"https://pith.science/paper/OXS46RLF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1907.13528&json=true","fetch_graph":"https://pith.science/api/pith-number/OXS46RLFZNRL3CTHKYB6O43EA4/graph.json","fetch_events":"https://pith.science/api/pith-number/OXS46RLFZNRL3CTHKYB6O43EA4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OXS46RLFZNRL3CTHKYB6O43EA4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OXS46RLFZNRL3CTHKYB6O43EA4/action/storage_attestation","attest_author":"https://pith.science/pith/OXS46RLFZNRL3CTHKYB6O43EA4/action/author_attestation","sign_citation":"https://pith.science/pith/OXS46RLFZNRL3CTHKYB6O43EA4/action/citation_signature","submit_replication":"https://pith.science/pith/OXS46RLFZNRL3CTHKYB6O43EA4/action/replication_record"}},"created_at":"2026-07-05T01:17:57.396071+00:00","updated_at":"2026-07-05T01:17:57.396071+00:00"}