{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:AQY45A6AB7R2FP7AKIILPS4EDJ","short_pith_number":"pith:AQY45A6A","schema_version":"1.0","canonical_sha256":"0431ce83c00fe3a2bfe05210b7cb841a77c9b65865dd9b0d2eacc3394d41251d","source":{"kind":"arxiv","id":"2505.03703","version":1},"attestation_state":"computed","paper":{"title":"Fill the Gap: Quantifying and Reducing the Modality Gap in Image-Text Representation Learning","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Fran\\c{c}ois Role, S\\'ebastien Meyer, Victor Amblard","submitted_at":"2025-05-06T17:24:41Z","abstract_excerpt":"Vision-language models (VLMs) allow to embed texts and images in a shared representation space. However, it has been shown that these models are subject to a modality gap phenomenon meaning there exists a clear separation between the embeddings from one modality and another in the embedding space. While this misalignment is detrimental for downstream tasks such as multimodal retrieval, multimodal clustering or zero-shot classification, etc. no generic and practical methods have so far been proposed to assess it precisely and even reduce it. We therefore propose novel measures and effective tec"},"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":"2505.03703","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2025-05-06T17:24:41Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"50e8600404ce61d4b6d158599e2dffc4642e95698ca5b91609c6efce2d986520","abstract_canon_sha256":"3042f67c4084d107fdcd29ade49dfff4e51c8e9aeb5f7aafb6f9ff6c860ddc34"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:59:20.792990Z","signature_b64":"2xYuc4zX0ERHSvDRVOnMwGQyX8xFtx5EVyKyCGjnWpA2zOKs0AfqXODDW97FSncMp2Ex8M0n8rgzCYE8M/SBCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0431ce83c00fe3a2bfe05210b7cb841a77c9b65865dd9b0d2eacc3394d41251d","last_reissued_at":"2026-07-05T10:59:20.792486Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:59:20.792486Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Fill the Gap: Quantifying and Reducing the Modality Gap in Image-Text Representation Learning","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Fran\\c{c}ois Role, S\\'ebastien Meyer, Victor Amblard","submitted_at":"2025-05-06T17:24:41Z","abstract_excerpt":"Vision-language models (VLMs) allow to embed texts and images in a shared representation space. However, it has been shown that these models are subject to a modality gap phenomenon meaning there exists a clear separation between the embeddings from one modality and another in the embedding space. While this misalignment is detrimental for downstream tasks such as multimodal retrieval, multimodal clustering or zero-shot classification, etc. no generic and practical methods have so far been proposed to assess it precisely and even reduce it. We therefore propose novel measures and effective tec"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.03703","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/2505.03703/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":"2505.03703","created_at":"2026-07-05T10:59:20.792549+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.03703v1","created_at":"2026-07-05T10:59:20.792549+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.03703","created_at":"2026-07-05T10:59:20.792549+00:00"},{"alias_kind":"pith_short_12","alias_value":"AQY45A6AB7R2","created_at":"2026-07-05T10:59:20.792549+00:00"},{"alias_kind":"pith_short_16","alias_value":"AQY45A6AB7R2FP7A","created_at":"2026-07-05T10:59:20.792549+00:00"},{"alias_kind":"pith_short_8","alias_value":"AQY45A6A","created_at":"2026-07-05T10:59:20.792549+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2510.05038","citing_title":"Guided Query Refinement: Multimodal Hybrid Retrieval with Test-Time Optimization","ref_index":27,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AQY45A6AB7R2FP7AKIILPS4EDJ","json":"https://pith.science/pith/AQY45A6AB7R2FP7AKIILPS4EDJ.json","graph_json":"https://pith.science/api/pith-number/AQY45A6AB7R2FP7AKIILPS4EDJ/graph.json","events_json":"https://pith.science/api/pith-number/AQY45A6AB7R2FP7AKIILPS4EDJ/events.json","paper":"https://pith.science/paper/AQY45A6A"},"agent_actions":{"view_html":"https://pith.science/pith/AQY45A6AB7R2FP7AKIILPS4EDJ","download_json":"https://pith.science/pith/AQY45A6AB7R2FP7AKIILPS4EDJ.json","view_paper":"https://pith.science/paper/AQY45A6A","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.03703&json=true","fetch_graph":"https://pith.science/api/pith-number/AQY45A6AB7R2FP7AKIILPS4EDJ/graph.json","fetch_events":"https://pith.science/api/pith-number/AQY45A6AB7R2FP7AKIILPS4EDJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AQY45A6AB7R2FP7AKIILPS4EDJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AQY45A6AB7R2FP7AKIILPS4EDJ/action/storage_attestation","attest_author":"https://pith.science/pith/AQY45A6AB7R2FP7AKIILPS4EDJ/action/author_attestation","sign_citation":"https://pith.science/pith/AQY45A6AB7R2FP7AKIILPS4EDJ/action/citation_signature","submit_replication":"https://pith.science/pith/AQY45A6AB7R2FP7AKIILPS4EDJ/action/replication_record"}},"created_at":"2026-07-05T10:59:20.792549+00:00","updated_at":"2026-07-05T10:59:20.792549+00:00"}