{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:LVXHARPWTONQXE6XOGUTNP6BN3","short_pith_number":"pith:LVXHARPW","schema_version":"1.0","canonical_sha256":"5d6e7045f69b9b0b93d771a936bfc16ee534b1fcf1d3d07adf06aea06bc60ea1","source":{"kind":"arxiv","id":"2407.16526","version":1},"attestation_state":"computed","paper":{"title":"Imperfect Vision Encoders: Efficient and Robust Tuning for Vision-Language Models","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.LG"],"primary_cat":"cs.CV","authors_text":"Aristeidis Panos, Daniel Olmeda Reino, Rahaf Aljundi, Richard E Turner","submitted_at":"2024-07-23T14:39:40Z","abstract_excerpt":"Vision language models (VLMs) demonstrate impressive capabilities in visual question answering and image captioning, acting as a crucial link between visual and language models. However, existing open-source VLMs heavily rely on pretrained and frozen vision encoders (such as CLIP). Despite CLIP's robustness across diverse domains, it still exhibits non-negligible image understanding errors. These errors propagate to the VLM responses, resulting in sub-optimal performance. In our work, we propose an efficient and robust method for updating vision encoders within VLMs. Our approach selectively a"},"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":"2407.16526","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CV","submitted_at":"2024-07-23T14:39:40Z","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"title_canon_sha256":"6d46389c2f5f0f0ef708cb3b87700ea20e36c7c267dbd9a5552fa7f46644999c","abstract_canon_sha256":"797afbede17c21eff2d9d734c8dd4ea3daedc184dbff51af1f9f4431c69be4fb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:47:34.705633Z","signature_b64":"Tlsb1kLN2LEDXXV69QhIvDKH3oXRevcY0l5V4nfbtLhAhGpvqNlXJpoA/qF+9gP73xFC9ZrnjA1Ktoj65gXqAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5d6e7045f69b9b0b93d771a936bfc16ee534b1fcf1d3d07adf06aea06bc60ea1","last_reissued_at":"2026-07-05T08:47:34.705113Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:47:34.705113Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Imperfect Vision Encoders: Efficient and Robust Tuning for Vision-Language Models","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.LG"],"primary_cat":"cs.CV","authors_text":"Aristeidis Panos, Daniel Olmeda Reino, Rahaf Aljundi, Richard E Turner","submitted_at":"2024-07-23T14:39:40Z","abstract_excerpt":"Vision language models (VLMs) demonstrate impressive capabilities in visual question answering and image captioning, acting as a crucial link between visual and language models. However, existing open-source VLMs heavily rely on pretrained and frozen vision encoders (such as CLIP). Despite CLIP's robustness across diverse domains, it still exhibits non-negligible image understanding errors. These errors propagate to the VLM responses, resulting in sub-optimal performance. In our work, we propose an efficient and robust method for updating vision encoders within VLMs. Our approach selectively a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.16526","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/2407.16526/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":"2407.16526","created_at":"2026-07-05T08:47:34.705173+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.16526v1","created_at":"2026-07-05T08:47:34.705173+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.16526","created_at":"2026-07-05T08:47:34.705173+00:00"},{"alias_kind":"pith_short_12","alias_value":"LVXHARPWTONQ","created_at":"2026-07-05T08:47:34.705173+00:00"},{"alias_kind":"pith_short_16","alias_value":"LVXHARPWTONQXE6X","created_at":"2026-07-05T08:47:34.705173+00:00"},{"alias_kind":"pith_short_8","alias_value":"LVXHARPW","created_at":"2026-07-05T08:47:34.705173+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.02904","citing_title":"Enhancing Sports Strategy with Video Analytics and Data Mining: Assessing the effectiveness of Multimodal LLMs in tennis video analysis","ref_index":30,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LVXHARPWTONQXE6XOGUTNP6BN3","json":"https://pith.science/pith/LVXHARPWTONQXE6XOGUTNP6BN3.json","graph_json":"https://pith.science/api/pith-number/LVXHARPWTONQXE6XOGUTNP6BN3/graph.json","events_json":"https://pith.science/api/pith-number/LVXHARPWTONQXE6XOGUTNP6BN3/events.json","paper":"https://pith.science/paper/LVXHARPW"},"agent_actions":{"view_html":"https://pith.science/pith/LVXHARPWTONQXE6XOGUTNP6BN3","download_json":"https://pith.science/pith/LVXHARPWTONQXE6XOGUTNP6BN3.json","view_paper":"https://pith.science/paper/LVXHARPW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.16526&json=true","fetch_graph":"https://pith.science/api/pith-number/LVXHARPWTONQXE6XOGUTNP6BN3/graph.json","fetch_events":"https://pith.science/api/pith-number/LVXHARPWTONQXE6XOGUTNP6BN3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LVXHARPWTONQXE6XOGUTNP6BN3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LVXHARPWTONQXE6XOGUTNP6BN3/action/storage_attestation","attest_author":"https://pith.science/pith/LVXHARPWTONQXE6XOGUTNP6BN3/action/author_attestation","sign_citation":"https://pith.science/pith/LVXHARPWTONQXE6XOGUTNP6BN3/action/citation_signature","submit_replication":"https://pith.science/pith/LVXHARPWTONQXE6XOGUTNP6BN3/action/replication_record"}},"created_at":"2026-07-05T08:47:34.705173+00:00","updated_at":"2026-07-05T08:47:34.705173+00:00"}