{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:U2QIH26AHR63XJK5IHEXGWL5CV","short_pith_number":"pith:U2QIH26A","schema_version":"1.0","canonical_sha256":"a6a083ebc03c7dbba55d41c973597d1569cafa620313240e5a88016bd0d16421","source":{"kind":"arxiv","id":"2401.11337","version":1},"attestation_state":"computed","paper":{"title":"Prompting Large Vision-Language Models for Compositional Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Junjie Hu, Ming Jiang, Timothy Ossowski","submitted_at":"2024-01-20T22:04:28Z","abstract_excerpt":"Vision-language models such as CLIP have shown impressive capabilities in encoding texts and images into aligned embeddings, enabling the retrieval of multimodal data in a shared embedding space. However, these embedding-based models still face challenges in effectively matching images and texts with similar visio-linguistic compositionality, as evidenced by their performance on the recent Winoground dataset. In this paper, we argue that this limitation stems from two factors: the use of single vector representations for complex multimodal data, and the absence of step-by-step reasoning in the"},"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":"2401.11337","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-01-20T22:04:28Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"3dbff1f6a6453d5f5f42ccff193cb395d4e0697b724a7c9e9ef796cdaf19b3d1","abstract_canon_sha256":"1541cbcb340a88cfd0c23890696a6e8b5d4e7e87704ad6131391649637335ad2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:35:49.272920Z","signature_b64":"O6jCy7tRkd5l/s5mMyMAgygzMF9TQkEHDjDCjA0XbrpFhaodw4OXXmCYfHyboVNqF8sDrNQ9T4kLWZwmX8plDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a6a083ebc03c7dbba55d41c973597d1569cafa620313240e5a88016bd0d16421","last_reissued_at":"2026-07-05T07:35:49.272495Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:35:49.272495Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Prompting Large Vision-Language Models for Compositional Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Junjie Hu, Ming Jiang, Timothy Ossowski","submitted_at":"2024-01-20T22:04:28Z","abstract_excerpt":"Vision-language models such as CLIP have shown impressive capabilities in encoding texts and images into aligned embeddings, enabling the retrieval of multimodal data in a shared embedding space. However, these embedding-based models still face challenges in effectively matching images and texts with similar visio-linguistic compositionality, as evidenced by their performance on the recent Winoground dataset. In this paper, we argue that this limitation stems from two factors: the use of single vector representations for complex multimodal data, and the absence of step-by-step reasoning in the"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.11337","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/2401.11337/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":"2401.11337","created_at":"2026-07-05T07:35:49.272561+00:00"},{"alias_kind":"arxiv_version","alias_value":"2401.11337v1","created_at":"2026-07-05T07:35:49.272561+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.11337","created_at":"2026-07-05T07:35:49.272561+00:00"},{"alias_kind":"pith_short_12","alias_value":"U2QIH26AHR63","created_at":"2026-07-05T07:35:49.272561+00:00"},{"alias_kind":"pith_short_16","alias_value":"U2QIH26AHR63XJK5","created_at":"2026-07-05T07:35:49.272561+00:00"},{"alias_kind":"pith_short_8","alias_value":"U2QIH26A","created_at":"2026-07-05T07:35:49.272561+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.00374","citing_title":"Learning to Compose: Revisiting Proxy Task Design for Zero-Shot Composed Image Retrieval","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2504.21850","citing_title":"Visual Compositional Tuning","ref_index":14,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/U2QIH26AHR63XJK5IHEXGWL5CV","json":"https://pith.science/pith/U2QIH26AHR63XJK5IHEXGWL5CV.json","graph_json":"https://pith.science/api/pith-number/U2QIH26AHR63XJK5IHEXGWL5CV/graph.json","events_json":"https://pith.science/api/pith-number/U2QIH26AHR63XJK5IHEXGWL5CV/events.json","paper":"https://pith.science/paper/U2QIH26A"},"agent_actions":{"view_html":"https://pith.science/pith/U2QIH26AHR63XJK5IHEXGWL5CV","download_json":"https://pith.science/pith/U2QIH26AHR63XJK5IHEXGWL5CV.json","view_paper":"https://pith.science/paper/U2QIH26A","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2401.11337&json=true","fetch_graph":"https://pith.science/api/pith-number/U2QIH26AHR63XJK5IHEXGWL5CV/graph.json","fetch_events":"https://pith.science/api/pith-number/U2QIH26AHR63XJK5IHEXGWL5CV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/U2QIH26AHR63XJK5IHEXGWL5CV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/U2QIH26AHR63XJK5IHEXGWL5CV/action/storage_attestation","attest_author":"https://pith.science/pith/U2QIH26AHR63XJK5IHEXGWL5CV/action/author_attestation","sign_citation":"https://pith.science/pith/U2QIH26AHR63XJK5IHEXGWL5CV/action/citation_signature","submit_replication":"https://pith.science/pith/U2QIH26AHR63XJK5IHEXGWL5CV/action/replication_record"}},"created_at":"2026-07-05T07:35:49.272561+00:00","updated_at":"2026-07-05T07:35:49.272561+00:00"}