{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:YOCS3DQBUQARVUPQJ52ONG5OH2","short_pith_number":"pith:YOCS3DQB","schema_version":"1.0","canonical_sha256":"c3852d8e01a4011ad1f04f74e69bae3e95da80c91153dbb8a16f9394c13549e8","source":{"kind":"arxiv","id":"1902.02671","version":2},"attestation_state":"computed","paper":{"title":"BERT and PALs: Projected Attention Layers for Efficient Adaptation in Multi-Task Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Asa Cooper Stickland, Iain Murray","submitted_at":"2019-02-07T15:05:46Z","abstract_excerpt":"Multi-task learning shares information between related tasks, sometimes reducing the number of parameters required. State-of-the-art results across multiple natural language understanding tasks in the GLUE benchmark have previously used transfer from a single large task: unsupervised pre-training with BERT, where a separate BERT model was fine-tuned for each task. We explore multi-task approaches that share a single BERT model with a small number of additional task-specific parameters. Using new adaptation modules, PALs or `projected attention layers', we match the performance of separately fi"},"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":"1902.02671","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-02-07T15:05:46Z","cross_cats_sorted":["cs.CL","stat.ML"],"title_canon_sha256":"031d8b2406613113338fa98f8a665aae25d08f1a4c4583f5fa949e670e83a9b2","abstract_canon_sha256":"76238d49e74e22ef19059509d285c919958d65dfd7aa0b48a75495adff4d9e93"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:46:09.143573Z","signature_b64":"6BlyFCCXEJtqsZz5EXkHoKLr8yeioTiFgSm7Nol1+4Ko4INjwk8EBbaU2RFISO4vZTCtNizfIjsfFTbF/MiZDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c3852d8e01a4011ad1f04f74e69bae3e95da80c91153dbb8a16f9394c13549e8","last_reissued_at":"2026-05-17T23:46:09.143162Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:46:09.143162Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"BERT and PALs: Projected Attention Layers for Efficient Adaptation in Multi-Task Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Asa Cooper Stickland, Iain Murray","submitted_at":"2019-02-07T15:05:46Z","abstract_excerpt":"Multi-task learning shares information between related tasks, sometimes reducing the number of parameters required. State-of-the-art results across multiple natural language understanding tasks in the GLUE benchmark have previously used transfer from a single large task: unsupervised pre-training with BERT, where a separate BERT model was fine-tuned for each task. We explore multi-task approaches that share a single BERT model with a small number of additional task-specific parameters. Using new adaptation modules, PALs or `projected attention layers', we match the performance of separately fi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1902.02671","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":""},"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":"1902.02671","created_at":"2026-05-17T23:46:09.143219+00:00"},{"alias_kind":"arxiv_version","alias_value":"1902.02671v2","created_at":"2026-05-17T23:46:09.143219+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1902.02671","created_at":"2026-05-17T23:46:09.143219+00:00"},{"alias_kind":"pith_short_12","alias_value":"YOCS3DQBUQAR","created_at":"2026-05-18T12:33:33.725879+00:00"},{"alias_kind":"pith_short_16","alias_value":"YOCS3DQBUQARVUPQ","created_at":"2026-05-18T12:33:33.725879+00:00"},{"alias_kind":"pith_short_8","alias_value":"YOCS3DQB","created_at":"2026-05-18T12:33:33.725879+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"1907.05338","citing_title":"To Tune or Not To Tune? How About the Best of Both Worlds?","ref_index":22,"is_internal_anchor":true},{"citing_arxiv_id":"2605.06477","citing_title":"GeoStack: A Framework for Quasi-Abelian Knowledge Composition in VLMs","ref_index":12,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YOCS3DQBUQARVUPQJ52ONG5OH2","json":"https://pith.science/pith/YOCS3DQBUQARVUPQJ52ONG5OH2.json","graph_json":"https://pith.science/api/pith-number/YOCS3DQBUQARVUPQJ52ONG5OH2/graph.json","events_json":"https://pith.science/api/pith-number/YOCS3DQBUQARVUPQJ52ONG5OH2/events.json","paper":"https://pith.science/paper/YOCS3DQB"},"agent_actions":{"view_html":"https://pith.science/pith/YOCS3DQBUQARVUPQJ52ONG5OH2","download_json":"https://pith.science/pith/YOCS3DQBUQARVUPQJ52ONG5OH2.json","view_paper":"https://pith.science/paper/YOCS3DQB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1902.02671&json=true","fetch_graph":"https://pith.science/api/pith-number/YOCS3DQBUQARVUPQJ52ONG5OH2/graph.json","fetch_events":"https://pith.science/api/pith-number/YOCS3DQBUQARVUPQJ52ONG5OH2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YOCS3DQBUQARVUPQJ52ONG5OH2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YOCS3DQBUQARVUPQJ52ONG5OH2/action/storage_attestation","attest_author":"https://pith.science/pith/YOCS3DQBUQARVUPQJ52ONG5OH2/action/author_attestation","sign_citation":"https://pith.science/pith/YOCS3DQBUQARVUPQJ52ONG5OH2/action/citation_signature","submit_replication":"https://pith.science/pith/YOCS3DQBUQARVUPQJ52ONG5OH2/action/replication_record"}},"created_at":"2026-05-17T23:46:09.143219+00:00","updated_at":"2026-05-17T23:46:09.143219+00:00"}