{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:5DATB7CFO4QZ2PVLEMAM2EOYLQ","short_pith_number":"pith:5DATB7CF","schema_version":"1.0","canonical_sha256":"e8c130fc4577219d3eab2300cd11d85c2cb7dcc722f7ff5be3c58b526ef1fdac","source":{"kind":"arxiv","id":"1906.05714","version":1},"attestation_state":"computed","paper":{"title":"A Multiscale Visualization of Attention in the Transformer Model","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.HC","authors_text":"Jesse Vig","submitted_at":"2019-06-12T15:45:26Z","abstract_excerpt":"The Transformer is a sequence model that forgoes traditional recurrent architectures in favor of a fully attention-based approach. Besides improving performance, an advantage of using attention is that it can also help to interpret a model by showing how the model assigns weight to different input elements. However, the multi-layer, multi-head attention mechanism in the Transformer model can be difficult to decipher. To make the model more accessible, we introduce an open-source tool that visualizes attention at multiple scales, each of which provides a unique perspective on the attention mech"},"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":"1906.05714","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.HC","submitted_at":"2019-06-12T15:45:26Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"33aa4d7d08e6613f33542bdff3aafd68e244ab41bda7c88546eac7368d267d64","abstract_canon_sha256":"00c449ce7c9121e4a95a1dc1e12d6f7e94a4fab63e422e1a76a6c8d4687f8bb4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:43:24.516850Z","signature_b64":"zTEofNDaGmjTagK2VpN6FmenneCX2lgDkRSq5J5BkrmvYRBlVoFUrHjVljhyAvhznFqVCXZyVHJ3/EcZ1s1yAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e8c130fc4577219d3eab2300cd11d85c2cb7dcc722f7ff5be3c58b526ef1fdac","last_reissued_at":"2026-05-17T23:43:24.516437Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:43:24.516437Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Multiscale Visualization of Attention in the Transformer Model","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.HC","authors_text":"Jesse Vig","submitted_at":"2019-06-12T15:45:26Z","abstract_excerpt":"The Transformer is a sequence model that forgoes traditional recurrent architectures in favor of a fully attention-based approach. Besides improving performance, an advantage of using attention is that it can also help to interpret a model by showing how the model assigns weight to different input elements. However, the multi-layer, multi-head attention mechanism in the Transformer model can be difficult to decipher. To make the model more accessible, we introduce an open-source tool that visualizes attention at multiple scales, each of which provides a unique perspective on the attention mech"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1906.05714","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":""},"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":"1906.05714","created_at":"2026-05-17T23:43:24.516503+00:00"},{"alias_kind":"arxiv_version","alias_value":"1906.05714v1","created_at":"2026-05-17T23:43:24.516503+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1906.05714","created_at":"2026-05-17T23:43:24.516503+00:00"},{"alias_kind":"pith_short_12","alias_value":"5DATB7CFO4QZ","created_at":"2026-05-18T12:33:10.108867+00:00"},{"alias_kind":"pith_short_16","alias_value":"5DATB7CFO4QZ2PVL","created_at":"2026-05-18T12:33:10.108867+00:00"},{"alias_kind":"pith_short_8","alias_value":"5DATB7CF","created_at":"2026-05-18T12:33:10.108867+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":7,"internal_anchor_count":4,"sample":[{"citing_arxiv_id":"1907.09117","citing_title":"Realistic Channel Models Pre-training","ref_index":11,"is_internal_anchor":true},{"citing_arxiv_id":"2311.04799","citing_title":"DA-Cramming: Enhancing Cost-Effective Language Model Pretraining with Dependency Agreement Integration","ref_index":28,"is_internal_anchor":true},{"citing_arxiv_id":"2409.10102","citing_title":"Trustworthiness in Retrieval-Augmented Generation Systems: A Survey","ref_index":121,"is_internal_anchor":true},{"citing_arxiv_id":"2602.16608","citing_title":"Explainable AI: Context-Aware Layer-Wise Integrated Gradients for Explaining Transformer Models","ref_index":36,"is_internal_anchor":true},{"citing_arxiv_id":"2009.14794","citing_title":"Rethinking Attention with Performers","ref_index":155,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06216","citing_title":"TIDE: Every Layer Knows the Token Beneath the Context","ref_index":70,"is_internal_anchor":false},{"citing_arxiv_id":"2209.11895","citing_title":"In-context Learning and Induction Heads","ref_index":8,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5DATB7CFO4QZ2PVLEMAM2EOYLQ","json":"https://pith.science/pith/5DATB7CFO4QZ2PVLEMAM2EOYLQ.json","graph_json":"https://pith.science/api/pith-number/5DATB7CFO4QZ2PVLEMAM2EOYLQ/graph.json","events_json":"https://pith.science/api/pith-number/5DATB7CFO4QZ2PVLEMAM2EOYLQ/events.json","paper":"https://pith.science/paper/5DATB7CF"},"agent_actions":{"view_html":"https://pith.science/pith/5DATB7CFO4QZ2PVLEMAM2EOYLQ","download_json":"https://pith.science/pith/5DATB7CFO4QZ2PVLEMAM2EOYLQ.json","view_paper":"https://pith.science/paper/5DATB7CF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1906.05714&json=true","fetch_graph":"https://pith.science/api/pith-number/5DATB7CFO4QZ2PVLEMAM2EOYLQ/graph.json","fetch_events":"https://pith.science/api/pith-number/5DATB7CFO4QZ2PVLEMAM2EOYLQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5DATB7CFO4QZ2PVLEMAM2EOYLQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5DATB7CFO4QZ2PVLEMAM2EOYLQ/action/storage_attestation","attest_author":"https://pith.science/pith/5DATB7CFO4QZ2PVLEMAM2EOYLQ/action/author_attestation","sign_citation":"https://pith.science/pith/5DATB7CFO4QZ2PVLEMAM2EOYLQ/action/citation_signature","submit_replication":"https://pith.science/pith/5DATB7CFO4QZ2PVLEMAM2EOYLQ/action/replication_record"}},"created_at":"2026-05-17T23:43:24.516503+00:00","updated_at":"2026-05-17T23:43:24.516503+00:00"}