{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:Y5KBYUFX4THZ25HJJRB5CGBLDV","short_pith_number":"pith:Y5KBYUFX","schema_version":"1.0","canonical_sha256":"c7541c50b7e4cf9d74e94c43d1182b1d7d203decc1b670a1438f1cdd61a33a7d","source":{"kind":"arxiv","id":"2006.15222","version":3},"attestation_state":"computed","paper":{"title":"BERTology Meets Biology: Interpreting Attention in Protein Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","q-bio.BM"],"primary_cat":"cs.CL","authors_text":"Ali Madani, Caiming Xiong, Jesse Vig, Lav R. Varshney, Nazneen Fatema Rajani, Richard Socher","submitted_at":"2020-06-26T21:50:17Z","abstract_excerpt":"Transformer architectures have proven to learn useful representations for protein classification and generation tasks. However, these representations present challenges in interpretability. In this work, we demonstrate a set of methods for analyzing protein Transformer models through the lens of attention. We show that attention: (1) captures the folding structure of proteins, connecting amino acids that are far apart in the underlying sequence, but spatially close in the three-dimensional structure, (2) targets binding sites, a key functional component of proteins, and (3) focuses on progress"},"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":"2006.15222","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-06-26T21:50:17Z","cross_cats_sorted":["cs.LG","q-bio.BM"],"title_canon_sha256":"bd70da7bbd691456827aed20e8ab4e034de75f44ca6db1c8f0ad5b20ca445e90","abstract_canon_sha256":"a602aa7d31a1669d682d71df65abf7a26d705c8024c867bbef846b68ea9ca5b3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:26:48.703243Z","signature_b64":"Hw4VtTISWe7SbZAtBHQGds0Tw9s8+h3/ADpTr7++L8Nd8RshjPnHBOxSE4XxprZQQpO1PCzRqEDd1WU9/VisAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c7541c50b7e4cf9d74e94c43d1182b1d7d203decc1b670a1438f1cdd61a33a7d","last_reissued_at":"2026-07-05T02:26:48.702760Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:26:48.702760Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"BERTology Meets Biology: Interpreting Attention in Protein Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","q-bio.BM"],"primary_cat":"cs.CL","authors_text":"Ali Madani, Caiming Xiong, Jesse Vig, Lav R. Varshney, Nazneen Fatema Rajani, Richard Socher","submitted_at":"2020-06-26T21:50:17Z","abstract_excerpt":"Transformer architectures have proven to learn useful representations for protein classification and generation tasks. However, these representations present challenges in interpretability. In this work, we demonstrate a set of methods for analyzing protein Transformer models through the lens of attention. We show that attention: (1) captures the folding structure of proteins, connecting amino acids that are far apart in the underlying sequence, but spatially close in the three-dimensional structure, (2) targets binding sites, a key functional component of proteins, and (3) focuses on progress"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2006.15222","kind":"arxiv","version":3},"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/2006.15222/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":"2006.15222","created_at":"2026-07-05T02:26:48.702820+00:00"},{"alias_kind":"arxiv_version","alias_value":"2006.15222v3","created_at":"2026-07-05T02:26:48.702820+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2006.15222","created_at":"2026-07-05T02:26:48.702820+00:00"},{"alias_kind":"pith_short_12","alias_value":"Y5KBYUFX4THZ","created_at":"2026-07-05T02:26:48.702820+00:00"},{"alias_kind":"pith_short_16","alias_value":"Y5KBYUFX4THZ25HJ","created_at":"2026-07-05T02:26:48.702820+00:00"},{"alias_kind":"pith_short_8","alias_value":"Y5KBYUFX","created_at":"2026-07-05T02:26:48.702820+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.16331","citing_title":"Retrieval and competition: how a protein foundation model starts a protein","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2310.00795","citing_title":"Recent Advances in Generative AI for Healthcare Applications","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2605.16331","citing_title":"Retrieval and competition: how a protein foundation model starts a protein","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2009.14794","citing_title":"Rethinking Attention with Performers","ref_index":157,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Y5KBYUFX4THZ25HJJRB5CGBLDV","json":"https://pith.science/pith/Y5KBYUFX4THZ25HJJRB5CGBLDV.json","graph_json":"https://pith.science/api/pith-number/Y5KBYUFX4THZ25HJJRB5CGBLDV/graph.json","events_json":"https://pith.science/api/pith-number/Y5KBYUFX4THZ25HJJRB5CGBLDV/events.json","paper":"https://pith.science/paper/Y5KBYUFX"},"agent_actions":{"view_html":"https://pith.science/pith/Y5KBYUFX4THZ25HJJRB5CGBLDV","download_json":"https://pith.science/pith/Y5KBYUFX4THZ25HJJRB5CGBLDV.json","view_paper":"https://pith.science/paper/Y5KBYUFX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2006.15222&json=true","fetch_graph":"https://pith.science/api/pith-number/Y5KBYUFX4THZ25HJJRB5CGBLDV/graph.json","fetch_events":"https://pith.science/api/pith-number/Y5KBYUFX4THZ25HJJRB5CGBLDV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Y5KBYUFX4THZ25HJJRB5CGBLDV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Y5KBYUFX4THZ25HJJRB5CGBLDV/action/storage_attestation","attest_author":"https://pith.science/pith/Y5KBYUFX4THZ25HJJRB5CGBLDV/action/author_attestation","sign_citation":"https://pith.science/pith/Y5KBYUFX4THZ25HJJRB5CGBLDV/action/citation_signature","submit_replication":"https://pith.science/pith/Y5KBYUFX4THZ25HJJRB5CGBLDV/action/replication_record"}},"created_at":"2026-07-05T02:26:48.702820+00:00","updated_at":"2026-07-05T02:26:48.702820+00:00"}