{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:4YBX6YG2HQTHIR4UQAYEH7S7G2","short_pith_number":"pith:4YBX6YG2","schema_version":"1.0","canonical_sha256":"e6037f60da3c26744794803043fe5f36a0543a61430ec4f0577f8a31968617d5","source":{"kind":"arxiv","id":"2109.04448","version":1},"attestation_state":"computed","paper":{"title":"Vision-and-Language or Vision-for-Language? On Cross-Modal Influence in Multimodal Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.CL","authors_text":"Desmond Elliott, Emanuele Bugliarello, Stella Frank","submitted_at":"2021-09-09T17:47:50Z","abstract_excerpt":"Pretrained vision-and-language BERTs aim to learn representations that combine information from both modalities. We propose a diagnostic method based on cross-modal input ablation to assess the extent to which these models actually integrate cross-modal information. This method involves ablating inputs from one modality, either entirely or selectively based on cross-modal grounding alignments, and evaluating the model prediction performance on the other modality. Model performance is measured by modality-specific tasks that mirror the model pretraining objectives (e.g. masked language modellin"},"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":"2109.04448","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-09-09T17:47:50Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"7e7deee9031edc718c4cce3b35ab06e3ad1559f30fd4ff6c8b83b3cd30bd3084","abstract_canon_sha256":"1f9c6697b3c483bd544a472db970c5316e874492da8b83655be8c5185a342e98"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:12:58.592607Z","signature_b64":"4KpoQzIh7pBdtv2Iv1hgYRcjbqFs5W85SIZqY4g3q/BiJWRded4QgWC9l/Mr40Xf0ds73DocwREZffzjx702Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e6037f60da3c26744794803043fe5f36a0543a61430ec4f0577f8a31968617d5","last_reissued_at":"2026-07-05T03:12:58.592137Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:12:58.592137Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Vision-and-Language or Vision-for-Language? On Cross-Modal Influence in Multimodal Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.CL","authors_text":"Desmond Elliott, Emanuele Bugliarello, Stella Frank","submitted_at":"2021-09-09T17:47:50Z","abstract_excerpt":"Pretrained vision-and-language BERTs aim to learn representations that combine information from both modalities. We propose a diagnostic method based on cross-modal input ablation to assess the extent to which these models actually integrate cross-modal information. This method involves ablating inputs from one modality, either entirely or selectively based on cross-modal grounding alignments, and evaluating the model prediction performance on the other modality. Model performance is measured by modality-specific tasks that mirror the model pretraining objectives (e.g. masked language modellin"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2109.04448","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/2109.04448/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":"2109.04448","created_at":"2026-07-05T03:12:58.592197+00:00"},{"alias_kind":"arxiv_version","alias_value":"2109.04448v1","created_at":"2026-07-05T03:12:58.592197+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2109.04448","created_at":"2026-07-05T03:12:58.592197+00:00"},{"alias_kind":"pith_short_12","alias_value":"4YBX6YG2HQTH","created_at":"2026-07-05T03:12:58.592197+00:00"},{"alias_kind":"pith_short_16","alias_value":"4YBX6YG2HQTHIR4U","created_at":"2026-07-05T03:12:58.592197+00:00"},{"alias_kind":"pith_short_8","alias_value":"4YBX6YG2","created_at":"2026-07-05T03:12:58.592197+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.17436","citing_title":"Medical Context Distorts Decisions in Clinical Vision Language Models","ref_index":7,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4YBX6YG2HQTHIR4UQAYEH7S7G2","json":"https://pith.science/pith/4YBX6YG2HQTHIR4UQAYEH7S7G2.json","graph_json":"https://pith.science/api/pith-number/4YBX6YG2HQTHIR4UQAYEH7S7G2/graph.json","events_json":"https://pith.science/api/pith-number/4YBX6YG2HQTHIR4UQAYEH7S7G2/events.json","paper":"https://pith.science/paper/4YBX6YG2"},"agent_actions":{"view_html":"https://pith.science/pith/4YBX6YG2HQTHIR4UQAYEH7S7G2","download_json":"https://pith.science/pith/4YBX6YG2HQTHIR4UQAYEH7S7G2.json","view_paper":"https://pith.science/paper/4YBX6YG2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2109.04448&json=true","fetch_graph":"https://pith.science/api/pith-number/4YBX6YG2HQTHIR4UQAYEH7S7G2/graph.json","fetch_events":"https://pith.science/api/pith-number/4YBX6YG2HQTHIR4UQAYEH7S7G2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4YBX6YG2HQTHIR4UQAYEH7S7G2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4YBX6YG2HQTHIR4UQAYEH7S7G2/action/storage_attestation","attest_author":"https://pith.science/pith/4YBX6YG2HQTHIR4UQAYEH7S7G2/action/author_attestation","sign_citation":"https://pith.science/pith/4YBX6YG2HQTHIR4UQAYEH7S7G2/action/citation_signature","submit_replication":"https://pith.science/pith/4YBX6YG2HQTHIR4UQAYEH7S7G2/action/replication_record"}},"created_at":"2026-07-05T03:12:58.592197+00:00","updated_at":"2026-07-05T03:12:58.592197+00:00"}