{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:4WN7BXYW7BOPIBZUQWO6AXTW3C","short_pith_number":"pith:4WN7BXYW","schema_version":"1.0","canonical_sha256":"e59bf0df16f85cf40734859de05e76d8a0fb0eddd6b359b61177f445e7f1c914","source":{"kind":"arxiv","id":"1906.02715","version":2},"attestation_state":"computed","paper":{"title":"Visualizing and Measuring the Geometry of BERT","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Adam Pearce, Andy Coenen, Ann Yuan, Been Kim, Emily Reif, Fernanda Vi\\'egas, Martin Wattenberg","submitted_at":"2019-06-06T17:33:22Z","abstract_excerpt":"Transformer architectures show significant promise for natural language processing. Given that a single pretrained model can be fine-tuned to perform well on many different tasks, these networks appear to extract generally useful linguistic features. A natural question is how such networks represent this information internally. This paper describes qualitative and quantitative investigations of one particularly effective model, BERT. At a high level, linguistic features seem to be represented in separate semantic and syntactic subspaces. We find evidence of a fine-grained geometric representat"},"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":"1906.02715","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-06-06T17:33:22Z","cross_cats_sorted":["cs.CL","stat.ML"],"title_canon_sha256":"e6d34ce35c8cc2c648a3f1520eff532d2261c6cd252deb9da2c875053d417bf1","abstract_canon_sha256":"90d58df5fb9e9a7d3fc3c672ce1ca75889c887e333d0e313494535c44581830d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:15:06.130345Z","signature_b64":"o1lrP1HcrIrsbFP03OwdXzhkm7YMXP00OuhMukqsJLumf6ybG+uMdrxKx4x82H7+SHEKoz0iaEbD8COSm3r1Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e59bf0df16f85cf40734859de05e76d8a0fb0eddd6b359b61177f445e7f1c914","last_reissued_at":"2026-07-05T00:15:06.129865Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:15:06.129865Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Visualizing and Measuring the Geometry of BERT","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Adam Pearce, Andy Coenen, Ann Yuan, Been Kim, Emily Reif, Fernanda Vi\\'egas, Martin Wattenberg","submitted_at":"2019-06-06T17:33:22Z","abstract_excerpt":"Transformer architectures show significant promise for natural language processing. Given that a single pretrained model can be fine-tuned to perform well on many different tasks, these networks appear to extract generally useful linguistic features. A natural question is how such networks represent this information internally. This paper describes qualitative and quantitative investigations of one particularly effective model, BERT. At a high level, linguistic features seem to be represented in separate semantic and syntactic subspaces. We find evidence of a fine-grained geometric representat"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1906.02715","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/1906.02715/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":"1906.02715","created_at":"2026-07-05T00:15:06.129924+00:00"},{"alias_kind":"arxiv_version","alias_value":"1906.02715v2","created_at":"2026-07-05T00:15:06.129924+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1906.02715","created_at":"2026-07-05T00:15:06.129924+00:00"},{"alias_kind":"pith_short_12","alias_value":"4WN7BXYW7BOP","created_at":"2026-07-05T00:15:06.129924+00:00"},{"alias_kind":"pith_short_16","alias_value":"4WN7BXYW7BOPIBZU","created_at":"2026-07-05T00:15:06.129924+00:00"},{"alias_kind":"pith_short_8","alias_value":"4WN7BXYW","created_at":"2026-07-05T00:15:06.129924+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2308.08089","citing_title":"DragNUWA: Fine-grained Control in Video Generation by Integrating Text, Image, and Trajectory","ref_index":162,"is_internal_anchor":false},{"citing_arxiv_id":"2012.06678","citing_title":"TabTransformer: Tabular Data Modeling Using Contextual Embeddings","ref_index":70,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01381","citing_title":"A framework for analyzing concept representations in neural models","ref_index":73,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4WN7BXYW7BOPIBZUQWO6AXTW3C","json":"https://pith.science/pith/4WN7BXYW7BOPIBZUQWO6AXTW3C.json","graph_json":"https://pith.science/api/pith-number/4WN7BXYW7BOPIBZUQWO6AXTW3C/graph.json","events_json":"https://pith.science/api/pith-number/4WN7BXYW7BOPIBZUQWO6AXTW3C/events.json","paper":"https://pith.science/paper/4WN7BXYW"},"agent_actions":{"view_html":"https://pith.science/pith/4WN7BXYW7BOPIBZUQWO6AXTW3C","download_json":"https://pith.science/pith/4WN7BXYW7BOPIBZUQWO6AXTW3C.json","view_paper":"https://pith.science/paper/4WN7BXYW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1906.02715&json=true","fetch_graph":"https://pith.science/api/pith-number/4WN7BXYW7BOPIBZUQWO6AXTW3C/graph.json","fetch_events":"https://pith.science/api/pith-number/4WN7BXYW7BOPIBZUQWO6AXTW3C/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4WN7BXYW7BOPIBZUQWO6AXTW3C/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4WN7BXYW7BOPIBZUQWO6AXTW3C/action/storage_attestation","attest_author":"https://pith.science/pith/4WN7BXYW7BOPIBZUQWO6AXTW3C/action/author_attestation","sign_citation":"https://pith.science/pith/4WN7BXYW7BOPIBZUQWO6AXTW3C/action/citation_signature","submit_replication":"https://pith.science/pith/4WN7BXYW7BOPIBZUQWO6AXTW3C/action/replication_record"}},"created_at":"2026-07-05T00:15:06.129924+00:00","updated_at":"2026-07-05T00:15:06.129924+00:00"}