{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2RZMIBPWDGO3P54ZR6WQ4CSWGU","short_pith_number":"pith:2RZMIBPW","schema_version":"1.0","canonical_sha256":"d472c405f6199db7f7998fad0e0a56351bc660c3940835047cb3fae57e17ef43","source":{"kind":"arxiv","id":"2502.14113","version":1},"attestation_state":"computed","paper":{"title":"Object-centric Binding in Contrastive Language-Image Pretraining","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Adriana Romero-Soriano, Florian Bordes, Michal Drozdzal, Pietro Astolfi, Rim Assouel","submitted_at":"2025-02-19T21:30:51Z","abstract_excerpt":"Recent advances in vision language models (VLM) have been driven by contrastive models such as CLIP, which learn to associate visual information with their corresponding text descriptions. However, these models have limitations in understanding complex compositional scenes involving multiple objects and their spatial relationships. To address these challenges, we propose a novel approach that diverges from commonly used strategies, which rely on the design of hard-negative augmentations. Instead, our work focuses on integrating inductive biases into pre-trained CLIP-like models to improve thei"},"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":"2502.14113","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2025-02-19T21:30:51Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"cacc7c5e1443e46935c59b5e689ce06def0ea59bbb0cdd47386be3d11db07ed3","abstract_canon_sha256":"59e725502414243af416740edb952032b95f167e86eed89f400d129b0b679cdf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:17:20.832755Z","signature_b64":"KVnHyjCpVgUcRI//ffVxCHrWm9Tw3Qok1E3iLcNJPJxRRcGW00sTHzNaJEWqJFfIsVErsdSiMnjz7/f71cTFCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d472c405f6199db7f7998fad0e0a56351bc660c3940835047cb3fae57e17ef43","last_reissued_at":"2026-07-05T10:17:20.832237Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:17:20.832237Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Object-centric Binding in Contrastive Language-Image Pretraining","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Adriana Romero-Soriano, Florian Bordes, Michal Drozdzal, Pietro Astolfi, Rim Assouel","submitted_at":"2025-02-19T21:30:51Z","abstract_excerpt":"Recent advances in vision language models (VLM) have been driven by contrastive models such as CLIP, which learn to associate visual information with their corresponding text descriptions. However, these models have limitations in understanding complex compositional scenes involving multiple objects and their spatial relationships. To address these challenges, we propose a novel approach that diverges from commonly used strategies, which rely on the design of hard-negative augmentations. Instead, our work focuses on integrating inductive biases into pre-trained CLIP-like models to improve thei"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.14113","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/2502.14113/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":"2502.14113","created_at":"2026-07-05T10:17:20.832301+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.14113v1","created_at":"2026-07-05T10:17:20.832301+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.14113","created_at":"2026-07-05T10:17:20.832301+00:00"},{"alias_kind":"pith_short_12","alias_value":"2RZMIBPWDGO3","created_at":"2026-07-05T10:17:20.832301+00:00"},{"alias_kind":"pith_short_16","alias_value":"2RZMIBPWDGO3P54Z","created_at":"2026-07-05T10:17:20.832301+00:00"},{"alias_kind":"pith_short_8","alias_value":"2RZMIBPW","created_at":"2026-07-05T10:17:20.832301+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.03976","citing_title":"Formalizing the Binding Problem","ref_index":92,"is_internal_anchor":false},{"citing_arxiv_id":"2605.25427","citing_title":"Binding Visual Features Point by Point","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2603.25722","citing_title":"No Hard Negatives Required: Concept Centric Learning Leads to Compositionality without Degrading Zero-shot Capabilities of Contrastive Models","ref_index":1,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2RZMIBPWDGO3P54ZR6WQ4CSWGU","json":"https://pith.science/pith/2RZMIBPWDGO3P54ZR6WQ4CSWGU.json","graph_json":"https://pith.science/api/pith-number/2RZMIBPWDGO3P54ZR6WQ4CSWGU/graph.json","events_json":"https://pith.science/api/pith-number/2RZMIBPWDGO3P54ZR6WQ4CSWGU/events.json","paper":"https://pith.science/paper/2RZMIBPW"},"agent_actions":{"view_html":"https://pith.science/pith/2RZMIBPWDGO3P54ZR6WQ4CSWGU","download_json":"https://pith.science/pith/2RZMIBPWDGO3P54ZR6WQ4CSWGU.json","view_paper":"https://pith.science/paper/2RZMIBPW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.14113&json=true","fetch_graph":"https://pith.science/api/pith-number/2RZMIBPWDGO3P54ZR6WQ4CSWGU/graph.json","fetch_events":"https://pith.science/api/pith-number/2RZMIBPWDGO3P54ZR6WQ4CSWGU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2RZMIBPWDGO3P54ZR6WQ4CSWGU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2RZMIBPWDGO3P54ZR6WQ4CSWGU/action/storage_attestation","attest_author":"https://pith.science/pith/2RZMIBPWDGO3P54ZR6WQ4CSWGU/action/author_attestation","sign_citation":"https://pith.science/pith/2RZMIBPWDGO3P54ZR6WQ4CSWGU/action/citation_signature","submit_replication":"https://pith.science/pith/2RZMIBPWDGO3P54ZR6WQ4CSWGU/action/replication_record"}},"created_at":"2026-07-05T10:17:20.832301+00:00","updated_at":"2026-07-05T10:17:20.832301+00:00"}