{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:LBV7BVV5DNO2U4TM3WBDAW375C","short_pith_number":"pith:LBV7BVV5","schema_version":"1.0","canonical_sha256":"586bf0d6bd1b5daa726cdd82305b7fe8ad72890196389839b3141e2a5b0b7499","source":{"kind":"arxiv","id":"2507.15807","version":2},"attestation_state":"computed","paper":{"title":"True Multimodal In-Context Learning Needs Attention to the Visual Context","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Daniel Cremers, Jianzhe Liu, Jindong Gu, Philip Torr, Shuo Chen, Volker Tresp, Yan Xia, Zhen Han","submitted_at":"2025-07-21T17:08:18Z","abstract_excerpt":"Multimodal Large Language Models (MLLMs), built on powerful language backbones, have enabled Multimodal In-Context Learning (MICL)-adapting to new tasks from a few multimodal demonstrations consisting of images, questions, and answers. Despite showing noticeable improvement on standard vision-language datasets, current MLLMs struggle to leverage visual information in the demonstrations. Specifically, they tend to neglect visual cues and over-rely on textual patterns, leading to mere text imitation rather than genuine multimodal adaptation. This behavior makes MICL still unimodal and largely re"},"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":"2507.15807","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-07-21T17:08:18Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"8ed66211ed0a9f1ba8569bbfa74dafcbccf1cb0a3bf94dde5fa579dc2e7e3b48","abstract_canon_sha256":"7056d72d0a0e4bc38685171bc87b964078f439c63cbc9afddc9f34c2124b6b7f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:49:11.661493Z","signature_b64":"wxcPp54sp9mtxXzbxNqyXbTQH2C5QxjKhurOyE1LvbqmttvuK3lJABHRMmEJk6gbRhYNKdTSxCmyxXdQXI1SBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"586bf0d6bd1b5daa726cdd82305b7fe8ad72890196389839b3141e2a5b0b7499","last_reissued_at":"2026-07-05T11:49:11.660981Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:49:11.660981Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"True Multimodal In-Context Learning Needs Attention to the Visual Context","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Daniel Cremers, Jianzhe Liu, Jindong Gu, Philip Torr, Shuo Chen, Volker Tresp, Yan Xia, Zhen Han","submitted_at":"2025-07-21T17:08:18Z","abstract_excerpt":"Multimodal Large Language Models (MLLMs), built on powerful language backbones, have enabled Multimodal In-Context Learning (MICL)-adapting to new tasks from a few multimodal demonstrations consisting of images, questions, and answers. Despite showing noticeable improvement on standard vision-language datasets, current MLLMs struggle to leverage visual information in the demonstrations. Specifically, they tend to neglect visual cues and over-rely on textual patterns, leading to mere text imitation rather than genuine multimodal adaptation. This behavior makes MICL still unimodal and largely re"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.15807","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/2507.15807/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":"2507.15807","created_at":"2026-07-05T11:49:11.661042+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.15807v2","created_at":"2026-07-05T11:49:11.661042+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.15807","created_at":"2026-07-05T11:49:11.661042+00:00"},{"alias_kind":"pith_short_12","alias_value":"LBV7BVV5DNO2","created_at":"2026-07-05T11:49:11.661042+00:00"},{"alias_kind":"pith_short_16","alias_value":"LBV7BVV5DNO2U4TM","created_at":"2026-07-05T11:49:11.661042+00:00"},{"alias_kind":"pith_short_8","alias_value":"LBV7BVV5","created_at":"2026-07-05T11:49:11.661042+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.08034","citing_title":"Sci-Rho: A Multilingual Visually-Grounded Symbolic Benchmark for STEM Problems","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2604.13403","citing_title":"Why Multimodal In-Context Learning Lags Behind? Unveiling the Inner Mechanisms and Bottlenecks","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02378","citing_title":"Enhancing Multimodal In-Context Learning via Inductive-Deductive Reasoning","ref_index":9,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LBV7BVV5DNO2U4TM3WBDAW375C","json":"https://pith.science/pith/LBV7BVV5DNO2U4TM3WBDAW375C.json","graph_json":"https://pith.science/api/pith-number/LBV7BVV5DNO2U4TM3WBDAW375C/graph.json","events_json":"https://pith.science/api/pith-number/LBV7BVV5DNO2U4TM3WBDAW375C/events.json","paper":"https://pith.science/paper/LBV7BVV5"},"agent_actions":{"view_html":"https://pith.science/pith/LBV7BVV5DNO2U4TM3WBDAW375C","download_json":"https://pith.science/pith/LBV7BVV5DNO2U4TM3WBDAW375C.json","view_paper":"https://pith.science/paper/LBV7BVV5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.15807&json=true","fetch_graph":"https://pith.science/api/pith-number/LBV7BVV5DNO2U4TM3WBDAW375C/graph.json","fetch_events":"https://pith.science/api/pith-number/LBV7BVV5DNO2U4TM3WBDAW375C/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LBV7BVV5DNO2U4TM3WBDAW375C/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LBV7BVV5DNO2U4TM3WBDAW375C/action/storage_attestation","attest_author":"https://pith.science/pith/LBV7BVV5DNO2U4TM3WBDAW375C/action/author_attestation","sign_citation":"https://pith.science/pith/LBV7BVV5DNO2U4TM3WBDAW375C/action/citation_signature","submit_replication":"https://pith.science/pith/LBV7BVV5DNO2U4TM3WBDAW375C/action/replication_record"}},"created_at":"2026-07-05T11:49:11.661042+00:00","updated_at":"2026-07-05T11:49:11.661042+00:00"}