{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:322GRY6IYU54IDVLIV32WMUATL","short_pith_number":"pith:322GRY6I","schema_version":"1.0","canonical_sha256":"deb468e3c8c53bc40eab4577ab32809ae128a90ee6a2f4becbbe5ca0af42d619","source":{"kind":"arxiv","id":"2402.17510","version":2},"attestation_state":"computed","paper":{"title":"Demonstrating and Reducing Shortcuts in Vision-Language Representation Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Andrew Yates, Maarten de Rijke, Mariya Hendriksen, Maurits Bleeker","submitted_at":"2024-02-27T13:50:34Z","abstract_excerpt":"Vision-language models (VLMs) mainly rely on contrastive training to learn general-purpose representations of images and captions. We focus on the situation when one image is associated with several captions, each caption containing both information shared among all captions and unique information per caption about the scene depicted in the image. In such cases, it is unclear whether contrastive losses are sufficient for learning task-optimal representations that contain all the information provided by the captions or whether the contrastive learning setup encourages the learning of a simple s"},"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":"2402.17510","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-02-27T13:50:34Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"4a234d7aa6ef052eff05b74fb4edaa577f69781649f7e3a72b3ae4940a573b7d","abstract_canon_sha256":"7d541924b08e67da12710781ced405cc6fc20f9df4ce74384ea0687666e964fd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:50:54.195921Z","signature_b64":"ctcYdyuy+AbLH6d8+r0i591woZef/oK2mS/0APal526dPEH9G+GWSuL91EThwkypTibUXN0pCmP9GX3EULBrDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"deb468e3c8c53bc40eab4577ab32809ae128a90ee6a2f4becbbe5ca0af42d619","last_reissued_at":"2026-07-05T08:50:54.195509Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:50:54.195509Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Demonstrating and Reducing Shortcuts in Vision-Language Representation Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Andrew Yates, Maarten de Rijke, Mariya Hendriksen, Maurits Bleeker","submitted_at":"2024-02-27T13:50:34Z","abstract_excerpt":"Vision-language models (VLMs) mainly rely on contrastive training to learn general-purpose representations of images and captions. We focus on the situation when one image is associated with several captions, each caption containing both information shared among all captions and unique information per caption about the scene depicted in the image. In such cases, it is unclear whether contrastive losses are sufficient for learning task-optimal representations that contain all the information provided by the captions or whether the contrastive learning setup encourages the learning of a simple s"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.17510","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/2402.17510/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":"2402.17510","created_at":"2026-07-05T08:50:54.195562+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.17510v2","created_at":"2026-07-05T08:50:54.195562+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.17510","created_at":"2026-07-05T08:50:54.195562+00:00"},{"alias_kind":"pith_short_12","alias_value":"322GRY6IYU54","created_at":"2026-07-05T08:50:54.195562+00:00"},{"alias_kind":"pith_short_16","alias_value":"322GRY6IYU54IDVL","created_at":"2026-07-05T08:50:54.195562+00:00"},{"alias_kind":"pith_short_8","alias_value":"322GRY6I","created_at":"2026-07-05T08:50:54.195562+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2508.19652","citing_title":"Self-Rewarding Vision-Language Model via Reasoning Decomposition","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2604.05117","citing_title":"Watch Before You Answer: Learning from Visually Grounded Post-Training","ref_index":8,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/322GRY6IYU54IDVLIV32WMUATL","json":"https://pith.science/pith/322GRY6IYU54IDVLIV32WMUATL.json","graph_json":"https://pith.science/api/pith-number/322GRY6IYU54IDVLIV32WMUATL/graph.json","events_json":"https://pith.science/api/pith-number/322GRY6IYU54IDVLIV32WMUATL/events.json","paper":"https://pith.science/paper/322GRY6I"},"agent_actions":{"view_html":"https://pith.science/pith/322GRY6IYU54IDVLIV32WMUATL","download_json":"https://pith.science/pith/322GRY6IYU54IDVLIV32WMUATL.json","view_paper":"https://pith.science/paper/322GRY6I","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.17510&json=true","fetch_graph":"https://pith.science/api/pith-number/322GRY6IYU54IDVLIV32WMUATL/graph.json","fetch_events":"https://pith.science/api/pith-number/322GRY6IYU54IDVLIV32WMUATL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/322GRY6IYU54IDVLIV32WMUATL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/322GRY6IYU54IDVLIV32WMUATL/action/storage_attestation","attest_author":"https://pith.science/pith/322GRY6IYU54IDVLIV32WMUATL/action/author_attestation","sign_citation":"https://pith.science/pith/322GRY6IYU54IDVLIV32WMUATL/action/citation_signature","submit_replication":"https://pith.science/pith/322GRY6IYU54IDVLIV32WMUATL/action/replication_record"}},"created_at":"2026-07-05T08:50:54.195562+00:00","updated_at":"2026-07-05T08:50:54.195562+00:00"}