{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:SBBRV7SIK32J4KVY6WGFPMXKCA","short_pith_number":"pith:SBBRV7SI","schema_version":"1.0","canonical_sha256":"90431afe4856f49e2ab8f58c57b2ea101e95b7403e1cf68082f3c2e4f82636b1","source":{"kind":"arxiv","id":"2404.07983","version":3},"attestation_state":"computed","paper":{"title":"Two Effects, One Trigger: On the Modality Gap, Object Bias, and Information Imbalance in Contrastive Vision-Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"David T. Hoffmann, Max Argus, Simon Schrodi, Thomas Brox, Volker Fischer","submitted_at":"2024-04-11T17:58:06Z","abstract_excerpt":"Contrastive vision-language models (VLMs), like CLIP, have gained popularity for their versatile applicability to various downstream tasks. Despite their successes in some tasks, like zero-shot object recognition, they perform surprisingly poor on other tasks, like attribute recognition. Previous work has attributed these challenges to the modality gap, a separation of image and text in the shared representation space, and to a bias towards objects over other factors, such as attributes. In this analysis paper, we investigate both phenomena thoroughly. We evaluated off-the-shelf VLMs and while"},"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":"2404.07983","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-04-11T17:58:06Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"8cb4b905a7a052eea0e3469cea1e6d41ad9f3d465ed4b3808935b30ebabadae0","abstract_canon_sha256":"285e079dbddfada5e16ef5a2fed4c8f992a8e519a6ab8c316288a0acc09d9e85"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:49:43.938058Z","signature_b64":"ccjnjVHOAS3CpsZvYAAWsBhI3mQYZuAhFVvTXFhtc4eUpggPlSJTE281ZVf6pA9u8alxkZzabifDj7/+eKI9BA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"90431afe4856f49e2ab8f58c57b2ea101e95b7403e1cf68082f3c2e4f82636b1","last_reissued_at":"2026-07-05T10:49:43.937596Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:49:43.937596Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Two Effects, One Trigger: On the Modality Gap, Object Bias, and Information Imbalance in Contrastive Vision-Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"David T. Hoffmann, Max Argus, Simon Schrodi, Thomas Brox, Volker Fischer","submitted_at":"2024-04-11T17:58:06Z","abstract_excerpt":"Contrastive vision-language models (VLMs), like CLIP, have gained popularity for their versatile applicability to various downstream tasks. Despite their successes in some tasks, like zero-shot object recognition, they perform surprisingly poor on other tasks, like attribute recognition. Previous work has attributed these challenges to the modality gap, a separation of image and text in the shared representation space, and to a bias towards objects over other factors, such as attributes. In this analysis paper, we investigate both phenomena thoroughly. We evaluated off-the-shelf VLMs and while"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.07983","kind":"arxiv","version":3},"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/2404.07983/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":"2404.07983","created_at":"2026-07-05T10:49:43.937647+00:00"},{"alias_kind":"arxiv_version","alias_value":"2404.07983v3","created_at":"2026-07-05T10:49:43.937647+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.07983","created_at":"2026-07-05T10:49:43.937647+00:00"},{"alias_kind":"pith_short_12","alias_value":"SBBRV7SIK32J","created_at":"2026-07-05T10:49:43.937647+00:00"},{"alias_kind":"pith_short_16","alias_value":"SBBRV7SIK32J4KVY","created_at":"2026-07-05T10:49:43.937647+00:00"},{"alias_kind":"pith_short_8","alias_value":"SBBRV7SI","created_at":"2026-07-05T10:49:43.937647+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.17950","citing_title":"Plug-and-Adapt: Multimodal Coreference Resolution at First Sight with a Pretrained Alignment Model","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2606.04385","citing_title":"Geometry-Preserving Unsupervised Alignment for Heterogeneous Foundation Models","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30265","citing_title":"LoMo: Local Modality Substitution for Deeper Vision-Language Fusion","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11659","citing_title":"Reviving In-domain Fine-tuning Methods for Source-Free Cross-domain Few-shot Learning","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03456","citing_title":"VL-SAM-v3: Memory-Guided Visual Priors for Open-World Object Detection","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03456","citing_title":"VL-SAM-v3: Memory-Guided Visual Priors for Open-World Object Detection","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10039","citing_title":"Counting to Four is still a Chore for VLMs","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03456","citing_title":"VL-SAM-v3: Memory-Guided Visual Priors for Open-World Object Detection","ref_index":39,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/SBBRV7SIK32J4KVY6WGFPMXKCA","json":"https://pith.science/pith/SBBRV7SIK32J4KVY6WGFPMXKCA.json","graph_json":"https://pith.science/api/pith-number/SBBRV7SIK32J4KVY6WGFPMXKCA/graph.json","events_json":"https://pith.science/api/pith-number/SBBRV7SIK32J4KVY6WGFPMXKCA/events.json","paper":"https://pith.science/paper/SBBRV7SI"},"agent_actions":{"view_html":"https://pith.science/pith/SBBRV7SIK32J4KVY6WGFPMXKCA","download_json":"https://pith.science/pith/SBBRV7SIK32J4KVY6WGFPMXKCA.json","view_paper":"https://pith.science/paper/SBBRV7SI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2404.07983&json=true","fetch_graph":"https://pith.science/api/pith-number/SBBRV7SIK32J4KVY6WGFPMXKCA/graph.json","fetch_events":"https://pith.science/api/pith-number/SBBRV7SIK32J4KVY6WGFPMXKCA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SBBRV7SIK32J4KVY6WGFPMXKCA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SBBRV7SIK32J4KVY6WGFPMXKCA/action/storage_attestation","attest_author":"https://pith.science/pith/SBBRV7SIK32J4KVY6WGFPMXKCA/action/author_attestation","sign_citation":"https://pith.science/pith/SBBRV7SIK32J4KVY6WGFPMXKCA/action/citation_signature","submit_replication":"https://pith.science/pith/SBBRV7SIK32J4KVY6WGFPMXKCA/action/replication_record"}},"created_at":"2026-07-05T10:49:43.937647+00:00","updated_at":"2026-07-05T10:49:43.937647+00:00"}