{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:YXBEYDIUVNKHSBITKCX4S3UY4L","short_pith_number":"pith:YXBEYDIU","schema_version":"1.0","canonical_sha256":"c5c24c0d14ab5479051350afc96e98e2d3cbfa85cf1317149b723edf43394f05","source":{"kind":"arxiv","id":"2304.04947","version":2},"attestation_state":"computed","paper":{"title":"Conditional Adapters: Parameter-efficient Transfer Learning with Fast Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bo Li, Joshua Ainslie, Junwen Bai, Kenton Lee, Ming-Wei Chang, Nan Du, Siddhartha Brahma, Tao Lei, Vincent Y. Zhao, Yanqi Zhou, Yuexin Wu, Yu Zhang","submitted_at":"2023-04-11T03:17:37Z","abstract_excerpt":"We propose Conditional Adapter (CoDA), a parameter-efficient transfer learning method that also improves inference efficiency. CoDA generalizes beyond standard adapter approaches to enable a new way of balancing speed and accuracy using conditional computation. Starting with an existing dense pretrained model, CoDA adds sparse activation together with a small number of new parameters and a light-weight training phase. Our experiments demonstrate that the CoDA approach provides an unexpectedly efficient way to transfer knowledge. Across a variety of language, vision, and speech tasks, CoDA achi"},"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":"2304.04947","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-04-11T03:17:37Z","cross_cats_sorted":[],"title_canon_sha256":"88cfd18e36a68e51fb4c86c61b0c7531c7cdbb9df5e3924548d87076ea95def0","abstract_canon_sha256":"a27271fc563bedc13b951dde6eeef3c288631966cdf9748899ca851749b81cf4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:16:40.107909Z","signature_b64":"et0iEwVPAfXX4Xw7f9mYcrDWEGD0MvwAvMNFxL4yhY8mtfxX28508siBSPVdLxpyuuFXV9z7/fNZ3Crc9vVuBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c5c24c0d14ab5479051350afc96e98e2d3cbfa85cf1317149b723edf43394f05","last_reissued_at":"2026-07-05T07:16:40.107385Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:16:40.107385Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Conditional Adapters: Parameter-efficient Transfer Learning with Fast Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bo Li, Joshua Ainslie, Junwen Bai, Kenton Lee, Ming-Wei Chang, Nan Du, Siddhartha Brahma, Tao Lei, Vincent Y. Zhao, Yanqi Zhou, Yuexin Wu, Yu Zhang","submitted_at":"2023-04-11T03:17:37Z","abstract_excerpt":"We propose Conditional Adapter (CoDA), a parameter-efficient transfer learning method that also improves inference efficiency. CoDA generalizes beyond standard adapter approaches to enable a new way of balancing speed and accuracy using conditional computation. Starting with an existing dense pretrained model, CoDA adds sparse activation together with a small number of new parameters and a light-weight training phase. Our experiments demonstrate that the CoDA approach provides an unexpectedly efficient way to transfer knowledge. Across a variety of language, vision, and speech tasks, CoDA achi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2304.04947","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/2304.04947/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":"2304.04947","created_at":"2026-07-05T07:16:40.107451+00:00"},{"alias_kind":"arxiv_version","alias_value":"2304.04947v2","created_at":"2026-07-05T07:16:40.107451+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2304.04947","created_at":"2026-07-05T07:16:40.107451+00:00"},{"alias_kind":"pith_short_12","alias_value":"YXBEYDIUVNKH","created_at":"2026-07-05T07:16:40.107451+00:00"},{"alias_kind":"pith_short_16","alias_value":"YXBEYDIUVNKHSBIT","created_at":"2026-07-05T07:16:40.107451+00:00"},{"alias_kind":"pith_short_8","alias_value":"YXBEYDIU","created_at":"2026-07-05T07:16:40.107451+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2403.14608","citing_title":"Parameter-Efficient Fine-Tuning for Large Models: A Comprehensive Survey","ref_index":34,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YXBEYDIUVNKHSBITKCX4S3UY4L","json":"https://pith.science/pith/YXBEYDIUVNKHSBITKCX4S3UY4L.json","graph_json":"https://pith.science/api/pith-number/YXBEYDIUVNKHSBITKCX4S3UY4L/graph.json","events_json":"https://pith.science/api/pith-number/YXBEYDIUVNKHSBITKCX4S3UY4L/events.json","paper":"https://pith.science/paper/YXBEYDIU"},"agent_actions":{"view_html":"https://pith.science/pith/YXBEYDIUVNKHSBITKCX4S3UY4L","download_json":"https://pith.science/pith/YXBEYDIUVNKHSBITKCX4S3UY4L.json","view_paper":"https://pith.science/paper/YXBEYDIU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2304.04947&json=true","fetch_graph":"https://pith.science/api/pith-number/YXBEYDIUVNKHSBITKCX4S3UY4L/graph.json","fetch_events":"https://pith.science/api/pith-number/YXBEYDIUVNKHSBITKCX4S3UY4L/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YXBEYDIUVNKHSBITKCX4S3UY4L/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YXBEYDIUVNKHSBITKCX4S3UY4L/action/storage_attestation","attest_author":"https://pith.science/pith/YXBEYDIUVNKHSBITKCX4S3UY4L/action/author_attestation","sign_citation":"https://pith.science/pith/YXBEYDIUVNKHSBITKCX4S3UY4L/action/citation_signature","submit_replication":"https://pith.science/pith/YXBEYDIUVNKHSBITKCX4S3UY4L/action/replication_record"}},"created_at":"2026-07-05T07:16:40.107451+00:00","updated_at":"2026-07-05T07:16:40.107451+00:00"}