{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:GHA3235MKKMZFHKY4JLCEHVQSY","short_pith_number":"pith:GHA3235M","schema_version":"1.0","canonical_sha256":"31c1bd6fac5299929d58e256221eb0960e211613f2e75247b6cca5a63ae38c60","source":{"kind":"arxiv","id":"2503.06670","version":1},"attestation_state":"computed","paper":{"title":"Attention, Please! PixelSHAP Reveals What Vision-Language Models Actually Focus On","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Roni Goldshmidt","submitted_at":"2025-03-09T15:43:55Z","abstract_excerpt":"Interpretability in Vision-Language Models (VLMs) is crucial for trust, debugging, and decision-making in high-stakes applications. We introduce PixelSHAP, a model-agnostic framework extending Shapley-based analysis to structured visual entities. Unlike previous methods focusing on text prompts, PixelSHAP applies to vision-based reasoning by systematically perturbing image objects and quantifying their influence on a VLM's response. PixelSHAP requires no model internals, operating solely on input-output pairs, making it compatible with open-source and commercial models. It supports diverse emb"},"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":"2503.06670","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-03-09T15:43:55Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"dbd82a3ffa34318c1a2796070008625bde0625b65d2e134275e1566e190c8162","abstract_canon_sha256":"2a9902f1924dd0b49031142a5cc9871377e0dec8b60aea0b6650d961fa9c2e1f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:27:39.842212Z","signature_b64":"Uu5cU6Jznecek3l/a3r91DQyoeWo3C2ys1kHozNDN4ohdfiHewltzlO6DZ0orkxF0jQapVgt4ra6YMRTyg2XBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"31c1bd6fac5299929d58e256221eb0960e211613f2e75247b6cca5a63ae38c60","last_reissued_at":"2026-07-05T10:27:39.841284Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:27:39.841284Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Attention, Please! PixelSHAP Reveals What Vision-Language Models Actually Focus On","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Roni Goldshmidt","submitted_at":"2025-03-09T15:43:55Z","abstract_excerpt":"Interpretability in Vision-Language Models (VLMs) is crucial for trust, debugging, and decision-making in high-stakes applications. We introduce PixelSHAP, a model-agnostic framework extending Shapley-based analysis to structured visual entities. Unlike previous methods focusing on text prompts, PixelSHAP applies to vision-based reasoning by systematically perturbing image objects and quantifying their influence on a VLM's response. PixelSHAP requires no model internals, operating solely on input-output pairs, making it compatible with open-source and commercial models. It supports diverse emb"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.06670","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/2503.06670/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":"2503.06670","created_at":"2026-07-05T10:27:39.841431+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.06670v1","created_at":"2026-07-05T10:27:39.841431+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.06670","created_at":"2026-07-05T10:27:39.841431+00:00"},{"alias_kind":"pith_short_12","alias_value":"GHA3235MKKMZ","created_at":"2026-07-05T10:27:39.841431+00:00"},{"alias_kind":"pith_short_16","alias_value":"GHA3235MKKMZFHKY","created_at":"2026-07-05T10:27:39.841431+00:00"},{"alias_kind":"pith_short_8","alias_value":"GHA3235M","created_at":"2026-07-05T10:27:39.841431+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2506.18985","citing_title":"GLIMPSE: Holistic Cross-Modal Explainability for Large Vision-Language Models","ref_index":26,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GHA3235MKKMZFHKY4JLCEHVQSY","json":"https://pith.science/pith/GHA3235MKKMZFHKY4JLCEHVQSY.json","graph_json":"https://pith.science/api/pith-number/GHA3235MKKMZFHKY4JLCEHVQSY/graph.json","events_json":"https://pith.science/api/pith-number/GHA3235MKKMZFHKY4JLCEHVQSY/events.json","paper":"https://pith.science/paper/GHA3235M"},"agent_actions":{"view_html":"https://pith.science/pith/GHA3235MKKMZFHKY4JLCEHVQSY","download_json":"https://pith.science/pith/GHA3235MKKMZFHKY4JLCEHVQSY.json","view_paper":"https://pith.science/paper/GHA3235M","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.06670&json=true","fetch_graph":"https://pith.science/api/pith-number/GHA3235MKKMZFHKY4JLCEHVQSY/graph.json","fetch_events":"https://pith.science/api/pith-number/GHA3235MKKMZFHKY4JLCEHVQSY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GHA3235MKKMZFHKY4JLCEHVQSY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GHA3235MKKMZFHKY4JLCEHVQSY/action/storage_attestation","attest_author":"https://pith.science/pith/GHA3235MKKMZFHKY4JLCEHVQSY/action/author_attestation","sign_citation":"https://pith.science/pith/GHA3235MKKMZFHKY4JLCEHVQSY/action/citation_signature","submit_replication":"https://pith.science/pith/GHA3235MKKMZFHKY4JLCEHVQSY/action/replication_record"}},"created_at":"2026-07-05T10:27:39.841431+00:00","updated_at":"2026-07-05T10:27:39.841431+00:00"}