{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:Q25BOS5FEZMLX4FEAE6SZ7PF7R","short_pith_number":"pith:Q25BOS5F","schema_version":"1.0","canonical_sha256":"86ba174ba52658bbf0a4013d2cfde5fc74d92961cc5b60287b24dffc0e384d0f","source":{"kind":"arxiv","id":"2306.00310","version":1},"attestation_state":"computed","paper":{"title":"Prompt Algebra for Task Composition","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Alessandro Achille, Luca Zancato, Matthew Trager, Pramuditha Perera, Stefano Soatto","submitted_at":"2023-06-01T03:20:54Z","abstract_excerpt":"We investigate whether prompts learned independently for different tasks can be later combined through prompt algebra to obtain a model that supports composition of tasks. We consider Visual Language Models (VLM) with prompt tuning as our base classifier and formally define the notion of prompt algebra. We propose constrained prompt tuning to improve performance of the composite classifier. In the proposed scheme, prompts are constrained to appear in the lower dimensional subspace spanned by the basis vectors of the pre-trained vocabulary. Further regularization is added to ensure that the lea"},"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":"2306.00310","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2023-06-01T03:20:54Z","cross_cats_sorted":[],"title_canon_sha256":"f52c957889aa6e4e821fc74d9e78d37938239603bf9b7c0c20852d2fd31d1e1c","abstract_canon_sha256":"cf4f79ee88b29db0676c8be90b2f49c28043e292b45c40ff2c31cff6f429d159"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:16:24.766886Z","signature_b64":"qzPdaIITH6KPG7nLZrrOaElXU7MtxMvXQ0Ahca/Mod4VTkHc+KBqtPnfkH3WuTu94cYBEig+aSrbRHsaNT9JCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"86ba174ba52658bbf0a4013d2cfde5fc74d92961cc5b60287b24dffc0e384d0f","last_reissued_at":"2026-07-05T06:16:24.766467Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:16:24.766467Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Prompt Algebra for Task Composition","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Alessandro Achille, Luca Zancato, Matthew Trager, Pramuditha Perera, Stefano Soatto","submitted_at":"2023-06-01T03:20:54Z","abstract_excerpt":"We investigate whether prompts learned independently for different tasks can be later combined through prompt algebra to obtain a model that supports composition of tasks. We consider Visual Language Models (VLM) with prompt tuning as our base classifier and formally define the notion of prompt algebra. We propose constrained prompt tuning to improve performance of the composite classifier. In the proposed scheme, prompts are constrained to appear in the lower dimensional subspace spanned by the basis vectors of the pre-trained vocabulary. Further regularization is added to ensure that the lea"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2306.00310","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/2306.00310/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":"2306.00310","created_at":"2026-07-05T06:16:24.766527+00:00"},{"alias_kind":"arxiv_version","alias_value":"2306.00310v1","created_at":"2026-07-05T06:16:24.766527+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2306.00310","created_at":"2026-07-05T06:16:24.766527+00:00"},{"alias_kind":"pith_short_12","alias_value":"Q25BOS5FEZML","created_at":"2026-07-05T06:16:24.766527+00:00"},{"alias_kind":"pith_short_16","alias_value":"Q25BOS5FEZMLX4FE","created_at":"2026-07-05T06:16:24.766527+00:00"},{"alias_kind":"pith_short_8","alias_value":"Q25BOS5F","created_at":"2026-07-05T06:16:24.766527+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2311.03658","citing_title":"The Linear Representation Hypothesis and the Geometry of Large Language Models","ref_index":22,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Q25BOS5FEZMLX4FEAE6SZ7PF7R","json":"https://pith.science/pith/Q25BOS5FEZMLX4FEAE6SZ7PF7R.json","graph_json":"https://pith.science/api/pith-number/Q25BOS5FEZMLX4FEAE6SZ7PF7R/graph.json","events_json":"https://pith.science/api/pith-number/Q25BOS5FEZMLX4FEAE6SZ7PF7R/events.json","paper":"https://pith.science/paper/Q25BOS5F"},"agent_actions":{"view_html":"https://pith.science/pith/Q25BOS5FEZMLX4FEAE6SZ7PF7R","download_json":"https://pith.science/pith/Q25BOS5FEZMLX4FEAE6SZ7PF7R.json","view_paper":"https://pith.science/paper/Q25BOS5F","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2306.00310&json=true","fetch_graph":"https://pith.science/api/pith-number/Q25BOS5FEZMLX4FEAE6SZ7PF7R/graph.json","fetch_events":"https://pith.science/api/pith-number/Q25BOS5FEZMLX4FEAE6SZ7PF7R/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Q25BOS5FEZMLX4FEAE6SZ7PF7R/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Q25BOS5FEZMLX4FEAE6SZ7PF7R/action/storage_attestation","attest_author":"https://pith.science/pith/Q25BOS5FEZMLX4FEAE6SZ7PF7R/action/author_attestation","sign_citation":"https://pith.science/pith/Q25BOS5FEZMLX4FEAE6SZ7PF7R/action/citation_signature","submit_replication":"https://pith.science/pith/Q25BOS5FEZMLX4FEAE6SZ7PF7R/action/replication_record"}},"created_at":"2026-07-05T06:16:24.766527+00:00","updated_at":"2026-07-05T06:16:24.766527+00:00"}