{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:FQZ7EEIKY3LC3VFFDERBHP4PRF","short_pith_number":"pith:FQZ7EEIK","schema_version":"1.0","canonical_sha256":"2c33f2110ac6d62dd4a5192213bf8f89581c80a7629271913f6d9cc67a4f1d83","source":{"kind":"arxiv","id":"2502.12084","version":4},"attestation_state":"computed","paper":{"title":"VLM2-Bench: A Closer Look at How Well VLMs Implicitly Link Explicit Matching Visual Cues","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dongyu Yao, Jianshu Zhang, Paul Pu Liang, Renjie Pi, Yi R. Fung","submitted_at":"2025-02-17T17:57:50Z","abstract_excerpt":"Visually linking matching cues is a crucial ability in daily life, such as identifying the same person in multiple photos based on their cues, even without knowing who they are. Despite the extensive knowledge that vision-language models (VLMs) possess, it remains largely unexplored whether they are capable of performing this fundamental task. To address this, we introduce \\textbf{VLM2-Bench}, a benchmark designed to assess whether VLMs can Visually Link Matching cues, with 9 subtasks and over 3,000 test cases. Comprehensive evaluation across twelve VLMs, along with further analysis of various"},"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":"2502.12084","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-02-17T17:57:50Z","cross_cats_sorted":[],"title_canon_sha256":"ed740f3f7da9d994aab793e7a9bcc4250a84ed4c6dd343119301b9279d935375","abstract_canon_sha256":"c390bd8dd034259da5be4b3549665e8d2ee037130613e235959bbe7a74feaa6f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:30:41.580296Z","signature_b64":"HCuPaUhj6eDNBoUTYAk8qNo1uuM7kzKcpGgk3QxE9ynPt/+/VGo74Jh2ZKF9y1hiVDSl/bsLCLUvscYFmYxGAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2c33f2110ac6d62dd4a5192213bf8f89581c80a7629271913f6d9cc67a4f1d83","last_reissued_at":"2026-07-05T11:30:41.579735Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:30:41.579735Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"VLM2-Bench: A Closer Look at How Well VLMs Implicitly Link Explicit Matching Visual Cues","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dongyu Yao, Jianshu Zhang, Paul Pu Liang, Renjie Pi, Yi R. Fung","submitted_at":"2025-02-17T17:57:50Z","abstract_excerpt":"Visually linking matching cues is a crucial ability in daily life, such as identifying the same person in multiple photos based on their cues, even without knowing who they are. Despite the extensive knowledge that vision-language models (VLMs) possess, it remains largely unexplored whether they are capable of performing this fundamental task. To address this, we introduce \\textbf{VLM2-Bench}, a benchmark designed to assess whether VLMs can Visually Link Matching cues, with 9 subtasks and over 3,000 test cases. Comprehensive evaluation across twelve VLMs, along with further analysis of various"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.12084","kind":"arxiv","version":4},"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/2502.12084/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":"2502.12084","created_at":"2026-07-05T11:30:41.579796+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.12084v4","created_at":"2026-07-05T11:30:41.579796+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.12084","created_at":"2026-07-05T11:30:41.579796+00:00"},{"alias_kind":"pith_short_12","alias_value":"FQZ7EEIKY3LC","created_at":"2026-07-05T11:30:41.579796+00:00"},{"alias_kind":"pith_short_16","alias_value":"FQZ7EEIKY3LC3VFF","created_at":"2026-07-05T11:30:41.579796+00:00"},{"alias_kind":"pith_short_8","alias_value":"FQZ7EEIK","created_at":"2026-07-05T11:30:41.579796+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.21337","citing_title":"DataClaw0: Agentic Tailoring Multimodal Data from Raw Streams","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2601.21692","citing_title":"TCAP: Tri-Component Attention Profiling for Unsupervised Backdoor Detection in MLLM Fine-Tuning","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2508.21094","citing_title":"EMCompress: Video-LLMs with Endomorphic Multimodal Compression","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2604.22498","citing_title":"CGC: Compositional Grounded Contrast for Fine-Grained Multi-Image Understanding","ref_index":61,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FQZ7EEIKY3LC3VFFDERBHP4PRF","json":"https://pith.science/pith/FQZ7EEIKY3LC3VFFDERBHP4PRF.json","graph_json":"https://pith.science/api/pith-number/FQZ7EEIKY3LC3VFFDERBHP4PRF/graph.json","events_json":"https://pith.science/api/pith-number/FQZ7EEIKY3LC3VFFDERBHP4PRF/events.json","paper":"https://pith.science/paper/FQZ7EEIK"},"agent_actions":{"view_html":"https://pith.science/pith/FQZ7EEIKY3LC3VFFDERBHP4PRF","download_json":"https://pith.science/pith/FQZ7EEIKY3LC3VFFDERBHP4PRF.json","view_paper":"https://pith.science/paper/FQZ7EEIK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.12084&json=true","fetch_graph":"https://pith.science/api/pith-number/FQZ7EEIKY3LC3VFFDERBHP4PRF/graph.json","fetch_events":"https://pith.science/api/pith-number/FQZ7EEIKY3LC3VFFDERBHP4PRF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FQZ7EEIKY3LC3VFFDERBHP4PRF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FQZ7EEIKY3LC3VFFDERBHP4PRF/action/storage_attestation","attest_author":"https://pith.science/pith/FQZ7EEIKY3LC3VFFDERBHP4PRF/action/author_attestation","sign_citation":"https://pith.science/pith/FQZ7EEIKY3LC3VFFDERBHP4PRF/action/citation_signature","submit_replication":"https://pith.science/pith/FQZ7EEIKY3LC3VFFDERBHP4PRF/action/replication_record"}},"created_at":"2026-07-05T11:30:41.579796+00:00","updated_at":"2026-07-05T11:30:41.579796+00:00"}