{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:P4ZH5VYV6PYERKUSI3F6PZZV5C","short_pith_number":"pith:P4ZH5VYV","schema_version":"1.0","canonical_sha256":"7f327ed715f3f048aa9246cbe7e735e89d4a14c451cc6b722e4dc395617ba6ea","source":{"kind":"arxiv","id":"2603.19862","version":2},"attestation_state":"computed","paper":{"title":"IsoCLIP: Decomposing CLIP Projectors for Efficient Intra-modal Alignment","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Andrew D. Bagdanov, Bart{\\l}omiej Twardowski, Dipam Goswami, Joost Van De Weijer, Marco Mistretta, Simone Magistri","submitted_at":"2026-03-20T11:25:55Z","abstract_excerpt":"Vision-Language Models like CLIP are extensively used for inter-modal tasks which involve both visual and text modalities. However, when the individual modality encoders are applied to inherently intra-modal tasks like image-to-image retrieval, their performance suffers from the intra-modal misalignment. In this paper we study intra-modal misalignment in CLIP with a focus on the role of the projectors that map pre-projection image and text embeddings into the shared embedding space. By analyzing the form of the cosine similarity applied to projected features, and its interaction with the contr"},"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":"2603.19862","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2026-03-20T11:25:55Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"ec9472f3bcc7e82a91017c1c5a98c75a72cbe0fb06e850b2d7cf50d731779b0a","abstract_canon_sha256":"4744d7a7e205867f05722193f31397358b387226fc2aeea5ca280cb89b6a5b89"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-01T01:03:51.999136Z","signature_b64":"fi8TqvmwRIWVJMb434IrxlAejGx4xSX0TQJ571LXaZ3JVDNPLYXzfqkrljWrPoC56y+G1CdQVWGv8sWyZkL/BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7f327ed715f3f048aa9246cbe7e735e89d4a14c451cc6b722e4dc395617ba6ea","last_reissued_at":"2026-06-01T01:03:51.997946Z","signature_status":"signed_v1","first_computed_at":"2026-06-01T01:03:51.997946Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"IsoCLIP: Decomposing CLIP Projectors for Efficient Intra-modal Alignment","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Andrew D. Bagdanov, Bart{\\l}omiej Twardowski, Dipam Goswami, Joost Van De Weijer, Marco Mistretta, Simone Magistri","submitted_at":"2026-03-20T11:25:55Z","abstract_excerpt":"Vision-Language Models like CLIP are extensively used for inter-modal tasks which involve both visual and text modalities. However, when the individual modality encoders are applied to inherently intra-modal tasks like image-to-image retrieval, their performance suffers from the intra-modal misalignment. In this paper we study intra-modal misalignment in CLIP with a focus on the role of the projectors that map pre-projection image and text embeddings into the shared embedding space. By analyzing the form of the cosine similarity applied to projected features, and its interaction with the contr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2603.19862","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/2603.19862/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":"2603.19862","created_at":"2026-06-01T01:03:51.998115+00:00"},{"alias_kind":"arxiv_version","alias_value":"2603.19862v2","created_at":"2026-06-01T01:03:51.998115+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.19862","created_at":"2026-06-01T01:03:51.998115+00:00"},{"alias_kind":"pith_short_12","alias_value":"P4ZH5VYV6PYE","created_at":"2026-06-01T01:03:51.998115+00:00"},{"alias_kind":"pith_short_16","alias_value":"P4ZH5VYV6PYERKUS","created_at":"2026-06-01T01:03:51.998115+00:00"},{"alias_kind":"pith_short_8","alias_value":"P4ZH5VYV","created_at":"2026-06-01T01:03:51.998115+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/P4ZH5VYV6PYERKUSI3F6PZZV5C","json":"https://pith.science/pith/P4ZH5VYV6PYERKUSI3F6PZZV5C.json","graph_json":"https://pith.science/api/pith-number/P4ZH5VYV6PYERKUSI3F6PZZV5C/graph.json","events_json":"https://pith.science/api/pith-number/P4ZH5VYV6PYERKUSI3F6PZZV5C/events.json","paper":"https://pith.science/paper/P4ZH5VYV"},"agent_actions":{"view_html":"https://pith.science/pith/P4ZH5VYV6PYERKUSI3F6PZZV5C","download_json":"https://pith.science/pith/P4ZH5VYV6PYERKUSI3F6PZZV5C.json","view_paper":"https://pith.science/paper/P4ZH5VYV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2603.19862&json=true","fetch_graph":"https://pith.science/api/pith-number/P4ZH5VYV6PYERKUSI3F6PZZV5C/graph.json","fetch_events":"https://pith.science/api/pith-number/P4ZH5VYV6PYERKUSI3F6PZZV5C/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/P4ZH5VYV6PYERKUSI3F6PZZV5C/action/timestamp_anchor","attest_storage":"https://pith.science/pith/P4ZH5VYV6PYERKUSI3F6PZZV5C/action/storage_attestation","attest_author":"https://pith.science/pith/P4ZH5VYV6PYERKUSI3F6PZZV5C/action/author_attestation","sign_citation":"https://pith.science/pith/P4ZH5VYV6PYERKUSI3F6PZZV5C/action/citation_signature","submit_replication":"https://pith.science/pith/P4ZH5VYV6PYERKUSI3F6PZZV5C/action/replication_record"}},"created_at":"2026-06-01T01:03:51.998115+00:00","updated_at":"2026-06-01T01:03:51.998115+00:00"}