{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:PJ22PWZ6W43QCLJ6YMYBWTGTPO","short_pith_number":"pith:PJ22PWZ6","schema_version":"1.0","canonical_sha256":"7a75a7db3eb737012d3ec3301b4cd37bb92054a85011b3cdee9f8dadd3235b50","source":{"kind":"arxiv","id":"2204.03574","version":3},"attestation_state":"computed","paper":{"title":"Learning to Compose Soft Prompts for Compositional Zero-Shot Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.CV"],"primary_cat":"cs.LG","authors_text":"Nihal V. Nayak, Peilin Yu, Stephen H. Bach","submitted_at":"2022-04-07T16:51:12Z","abstract_excerpt":"We introduce compositional soft prompting (CSP), a parameter-efficient learning technique to improve the zero-shot compositionality of large-scale pretrained vision-language models (VLMs) like CLIP. We develop CSP for compositional zero-shot learning, the task of predicting unseen attribute-object compositions (e.g., old cat and young tiger). VLMs have a flexible text encoder that can represent arbitrary classes as natural language prompts but they often underperform task-specific architectures on the compositional zero-shot benchmark datasets. CSP treats the attributes and objects that define"},"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":"2204.03574","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-04-07T16:51:12Z","cross_cats_sorted":["cs.CL","cs.CV"],"title_canon_sha256":"9561920943d026f5c730a702d941aba5147ebe19c74bd1018006f0dbfb8a0392","abstract_canon_sha256":"1970d021c841ed1e325df754f069cb0f6b872f0151e393be716ba1eebb729a96"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:03:44.830697Z","signature_b64":"3stugxTurYXFBLF/YWQXnH7TmJtcE1MJRUDDgEqLbmWEfQ1fdA9m9OUeiWnlmq1zTd0/6zvlE0l7mET+cJYJAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7a75a7db3eb737012d3ec3301b4cd37bb92054a85011b3cdee9f8dadd3235b50","last_reissued_at":"2026-07-05T06:03:44.830244Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:03:44.830244Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning to Compose Soft Prompts for Compositional Zero-Shot Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.CV"],"primary_cat":"cs.LG","authors_text":"Nihal V. Nayak, Peilin Yu, Stephen H. Bach","submitted_at":"2022-04-07T16:51:12Z","abstract_excerpt":"We introduce compositional soft prompting (CSP), a parameter-efficient learning technique to improve the zero-shot compositionality of large-scale pretrained vision-language models (VLMs) like CLIP. We develop CSP for compositional zero-shot learning, the task of predicting unseen attribute-object compositions (e.g., old cat and young tiger). VLMs have a flexible text encoder that can represent arbitrary classes as natural language prompts but they often underperform task-specific architectures on the compositional zero-shot benchmark datasets. CSP treats the attributes and objects that define"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2204.03574","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/2204.03574/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":"2204.03574","created_at":"2026-07-05T06:03:44.830300+00:00"},{"alias_kind":"arxiv_version","alias_value":"2204.03574v3","created_at":"2026-07-05T06:03:44.830300+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2204.03574","created_at":"2026-07-05T06:03:44.830300+00:00"},{"alias_kind":"pith_short_12","alias_value":"PJ22PWZ6W43Q","created_at":"2026-07-05T06:03:44.830300+00:00"},{"alias_kind":"pith_short_16","alias_value":"PJ22PWZ6W43QCLJ6","created_at":"2026-07-05T06:03:44.830300+00:00"},{"alias_kind":"pith_short_8","alias_value":"PJ22PWZ6","created_at":"2026-07-05T06:03:44.830300+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2506.20986","citing_title":"EVA: Mixture-of-Experts Semantic Variant Alignment for Compositional Zero-Shot Learning","ref_index":31,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PJ22PWZ6W43QCLJ6YMYBWTGTPO","json":"https://pith.science/pith/PJ22PWZ6W43QCLJ6YMYBWTGTPO.json","graph_json":"https://pith.science/api/pith-number/PJ22PWZ6W43QCLJ6YMYBWTGTPO/graph.json","events_json":"https://pith.science/api/pith-number/PJ22PWZ6W43QCLJ6YMYBWTGTPO/events.json","paper":"https://pith.science/paper/PJ22PWZ6"},"agent_actions":{"view_html":"https://pith.science/pith/PJ22PWZ6W43QCLJ6YMYBWTGTPO","download_json":"https://pith.science/pith/PJ22PWZ6W43QCLJ6YMYBWTGTPO.json","view_paper":"https://pith.science/paper/PJ22PWZ6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2204.03574&json=true","fetch_graph":"https://pith.science/api/pith-number/PJ22PWZ6W43QCLJ6YMYBWTGTPO/graph.json","fetch_events":"https://pith.science/api/pith-number/PJ22PWZ6W43QCLJ6YMYBWTGTPO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PJ22PWZ6W43QCLJ6YMYBWTGTPO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PJ22PWZ6W43QCLJ6YMYBWTGTPO/action/storage_attestation","attest_author":"https://pith.science/pith/PJ22PWZ6W43QCLJ6YMYBWTGTPO/action/author_attestation","sign_citation":"https://pith.science/pith/PJ22PWZ6W43QCLJ6YMYBWTGTPO/action/citation_signature","submit_replication":"https://pith.science/pith/PJ22PWZ6W43QCLJ6YMYBWTGTPO/action/replication_record"}},"created_at":"2026-07-05T06:03:44.830300+00:00","updated_at":"2026-07-05T06:03:44.830300+00:00"}