{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:2XAI5CBNPP7NM27G6U26BO6L4N","short_pith_number":"pith:2XAI5CBN","schema_version":"1.0","canonical_sha256":"d5c08e882d7bfed66be6f535e0bbcbe35a73cf8607c27bf7987e81488838d22d","source":{"kind":"arxiv","id":"2104.06045","version":1},"attestation_state":"computed","paper":{"title":"Structural analysis of an all-purpose question answering model","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Fran\\c{c}ois Fleuret, Quentin Heinrich, Vincent Micheli, Wacim Belblidia","submitted_at":"2021-04-13T09:20:44Z","abstract_excerpt":"Attention is a key component of the now ubiquitous pre-trained language models. By learning to focus on relevant pieces of information, these Transformer-based architectures have proven capable of tackling several tasks at once and sometimes even surpass their single-task counterparts. To better understand this phenomenon, we conduct a structural analysis of a new all-purpose question answering model that we introduce. Surprisingly, this model retains single-task performance even in the absence of a strong transfer effect between tasks. Through attention head importance scoring, we observe tha"},"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":"2104.06045","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2021-04-13T09:20:44Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"d15dd883f5754f8886798c01dfa5a8ae15eaafb0849a0312a64c55fcd8bfeda5","abstract_canon_sha256":"37272b4c8cf712944aa16317ee476dd0115292f24556dde5761d000d68b0e7de"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:31:28.414775Z","signature_b64":"pwbGKMuoPE9lmvWdiB6bkmEN7txds5EiHc+6otKh77OWUggu5MpOjpgIqt+gT2BrbSxu+WGDv3fIdYWP4s7jCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d5c08e882d7bfed66be6f535e0bbcbe35a73cf8607c27bf7987e81488838d22d","last_reissued_at":"2026-07-05T02:31:28.414296Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:31:28.414296Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Structural analysis of an all-purpose question answering model","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Fran\\c{c}ois Fleuret, Quentin Heinrich, Vincent Micheli, Wacim Belblidia","submitted_at":"2021-04-13T09:20:44Z","abstract_excerpt":"Attention is a key component of the now ubiquitous pre-trained language models. By learning to focus on relevant pieces of information, these Transformer-based architectures have proven capable of tackling several tasks at once and sometimes even surpass their single-task counterparts. To better understand this phenomenon, we conduct a structural analysis of a new all-purpose question answering model that we introduce. Surprisingly, this model retains single-task performance even in the absence of a strong transfer effect between tasks. Through attention head importance scoring, we observe tha"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2104.06045","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/2104.06045/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":"2104.06045","created_at":"2026-07-05T02:31:28.414364+00:00"},{"alias_kind":"arxiv_version","alias_value":"2104.06045v1","created_at":"2026-07-05T02:31:28.414364+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2104.06045","created_at":"2026-07-05T02:31:28.414364+00:00"},{"alias_kind":"pith_short_12","alias_value":"2XAI5CBNPP7N","created_at":"2026-07-05T02:31:28.414364+00:00"},{"alias_kind":"pith_short_16","alias_value":"2XAI5CBNPP7NM27G","created_at":"2026-07-05T02:31:28.414364+00:00"},{"alias_kind":"pith_short_8","alias_value":"2XAI5CBN","created_at":"2026-07-05T02:31:28.414364+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2XAI5CBNPP7NM27G6U26BO6L4N","json":"https://pith.science/pith/2XAI5CBNPP7NM27G6U26BO6L4N.json","graph_json":"https://pith.science/api/pith-number/2XAI5CBNPP7NM27G6U26BO6L4N/graph.json","events_json":"https://pith.science/api/pith-number/2XAI5CBNPP7NM27G6U26BO6L4N/events.json","paper":"https://pith.science/paper/2XAI5CBN"},"agent_actions":{"view_html":"https://pith.science/pith/2XAI5CBNPP7NM27G6U26BO6L4N","download_json":"https://pith.science/pith/2XAI5CBNPP7NM27G6U26BO6L4N.json","view_paper":"https://pith.science/paper/2XAI5CBN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2104.06045&json=true","fetch_graph":"https://pith.science/api/pith-number/2XAI5CBNPP7NM27G6U26BO6L4N/graph.json","fetch_events":"https://pith.science/api/pith-number/2XAI5CBNPP7NM27G6U26BO6L4N/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2XAI5CBNPP7NM27G6U26BO6L4N/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2XAI5CBNPP7NM27G6U26BO6L4N/action/storage_attestation","attest_author":"https://pith.science/pith/2XAI5CBNPP7NM27G6U26BO6L4N/action/author_attestation","sign_citation":"https://pith.science/pith/2XAI5CBNPP7NM27G6U26BO6L4N/action/citation_signature","submit_replication":"https://pith.science/pith/2XAI5CBNPP7NM27G6U26BO6L4N/action/replication_record"}},"created_at":"2026-07-05T02:31:28.414364+00:00","updated_at":"2026-07-05T02:31:28.414364+00:00"}