{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:4RJMAJXCF7VT6PRQZNAAWLHCOF","short_pith_number":"pith:4RJMAJXC","schema_version":"1.0","canonical_sha256":"e452c026e22feb3f3e30cb400b2ce2716815b60a72364402a75c68bc5267d85d","source":{"kind":"arxiv","id":"2304.06712","version":2},"attestation_state":"computed","paper":{"title":"What does CLIP know about a red circle? Visual prompt engineering for VLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Aleksandar Shtedritski, Andrea Vedaldi, Christian Rupprecht","submitted_at":"2023-04-13T17:58:08Z","abstract_excerpt":"Large-scale Vision-Language Models, such as CLIP, learn powerful image-text representations that have found numerous applications, from zero-shot classification to text-to-image generation. Despite that, their capabilities for solving novel discriminative tasks via prompting fall behind those of large language models, such as GPT-3. Here we explore the idea of visual prompt engineering for solving computer vision tasks beyond classification by editing in image space instead of text. In particular, we discover an emergent ability of CLIP, where, by simply drawing a red circle around an object, "},"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":"2304.06712","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2023-04-13T17:58:08Z","cross_cats_sorted":[],"title_canon_sha256":"f92a5e1fc44f0f72a1ac40f872fd2011bedad54ae57f980e55578e8b6edd9f02","abstract_canon_sha256":"b8b6edeabc5bf5628bc9de649128b2f5bcbcf50cdcb8b0ae7331fd67f1568c71"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:42:21.077891Z","signature_b64":"SjcSAtntE6kzG24Rbd7j3mwY0lw4n8ezwj8+q/9CUd2peDXCytFTAJZsa4FZm/oN0QZHLHZQKSLpg+qtcGC1CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e452c026e22feb3f3e30cb400b2ce2716815b60a72364402a75c68bc5267d85d","last_reissued_at":"2026-07-05T06:42:21.077420Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:42:21.077420Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"What does CLIP know about a red circle? Visual prompt engineering for VLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Aleksandar Shtedritski, Andrea Vedaldi, Christian Rupprecht","submitted_at":"2023-04-13T17:58:08Z","abstract_excerpt":"Large-scale Vision-Language Models, such as CLIP, learn powerful image-text representations that have found numerous applications, from zero-shot classification to text-to-image generation. Despite that, their capabilities for solving novel discriminative tasks via prompting fall behind those of large language models, such as GPT-3. Here we explore the idea of visual prompt engineering for solving computer vision tasks beyond classification by editing in image space instead of text. In particular, we discover an emergent ability of CLIP, where, by simply drawing a red circle around an object, "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2304.06712","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/2304.06712/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":"2304.06712","created_at":"2026-07-05T06:42:21.077475+00:00"},{"alias_kind":"arxiv_version","alias_value":"2304.06712v2","created_at":"2026-07-05T06:42:21.077475+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2304.06712","created_at":"2026-07-05T06:42:21.077475+00:00"},{"alias_kind":"pith_short_12","alias_value":"4RJMAJXCF7VT","created_at":"2026-07-05T06:42:21.077475+00:00"},{"alias_kind":"pith_short_16","alias_value":"4RJMAJXCF7VT6PRQ","created_at":"2026-07-05T06:42:21.077475+00:00"},{"alias_kind":"pith_short_8","alias_value":"4RJMAJXC","created_at":"2026-07-05T06:42:21.077475+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2403.13805","citing_title":"RAR: Retrieving And Ranking Augmented MLLMs for Visual Recognition","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2309.17421","citing_title":"The Dawn of LMMs: Preliminary Explorations with GPT-4V(ision)","ref_index":117,"is_internal_anchor":false},{"citing_arxiv_id":"2404.12390","citing_title":"BLINK: Multimodal Large Language Models Can See but Not Perceive","ref_index":67,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4RJMAJXCF7VT6PRQZNAAWLHCOF","json":"https://pith.science/pith/4RJMAJXCF7VT6PRQZNAAWLHCOF.json","graph_json":"https://pith.science/api/pith-number/4RJMAJXCF7VT6PRQZNAAWLHCOF/graph.json","events_json":"https://pith.science/api/pith-number/4RJMAJXCF7VT6PRQZNAAWLHCOF/events.json","paper":"https://pith.science/paper/4RJMAJXC"},"agent_actions":{"view_html":"https://pith.science/pith/4RJMAJXCF7VT6PRQZNAAWLHCOF","download_json":"https://pith.science/pith/4RJMAJXCF7VT6PRQZNAAWLHCOF.json","view_paper":"https://pith.science/paper/4RJMAJXC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2304.06712&json=true","fetch_graph":"https://pith.science/api/pith-number/4RJMAJXCF7VT6PRQZNAAWLHCOF/graph.json","fetch_events":"https://pith.science/api/pith-number/4RJMAJXCF7VT6PRQZNAAWLHCOF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4RJMAJXCF7VT6PRQZNAAWLHCOF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4RJMAJXCF7VT6PRQZNAAWLHCOF/action/storage_attestation","attest_author":"https://pith.science/pith/4RJMAJXCF7VT6PRQZNAAWLHCOF/action/author_attestation","sign_citation":"https://pith.science/pith/4RJMAJXCF7VT6PRQZNAAWLHCOF/action/citation_signature","submit_replication":"https://pith.science/pith/4RJMAJXCF7VT6PRQZNAAWLHCOF/action/replication_record"}},"created_at":"2026-07-05T06:42:21.077475+00:00","updated_at":"2026-07-05T06:42:21.077475+00:00"}