{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:FKH6QOCRNRX2DDMSX2PGRZV4C2","short_pith_number":"pith:FKH6QOCR","schema_version":"1.0","canonical_sha256":"2a8fe838516c6fa18d92be9e68e6bc16abef30d10c5935a739ff0c67ed6a897a","source":{"kind":"arxiv","id":"2411.17451","version":2},"attestation_state":"computed","paper":{"title":"VL-RewardBench: A Challenging Benchmark for Vision-Language Generative Reward Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Bill Yuchen Lin, Chenxin An, Lei Li, Lingpeng Kong, Peiyi Wang, Qi Liu, Sujian Li, Tianyu Liu, Xuqing Yang, Yifan Song, Yuancheng Wei, Zhihui Xie","submitted_at":"2024-11-26T14:08:34Z","abstract_excerpt":"Vision-language generative reward models (VL-GenRMs) play a crucial role in aligning and evaluating multimodal AI systems, yet their own evaluation remains under-explored. Current assessment methods primarily rely on AI-annotated preference labels from traditional VL tasks, which can introduce biases and often fail to effectively challenge state-of-the-art models. To address these limitations, we introduce VL-RewardBench, a comprehensive benchmark spanning general multimodal queries, visual hallucination detection, and complex reasoning tasks. Through our AI-assisted annotation pipeline that c"},"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":"2411.17451","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-11-26T14:08:34Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"81ad3178422e98c0ab0da61205f0f6d6f87a85fcb08b9a6b7dd12ee5419d6d60","abstract_canon_sha256":"fc38dd20736758e627568e1c04bb50e26bd149410a67f3097bdf99611042058e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:14:00.196400Z","signature_b64":"MwUhnOikkbUlgcH87RTwRNCfBHpItabzr23S5uHo1aQujLwHNFQeYAygAl/4e++WmZbK/rkUrjttYLjHHz3ECA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2a8fe838516c6fa18d92be9e68e6bc16abef30d10c5935a739ff0c67ed6a897a","last_reissued_at":"2026-07-05T11:14:00.195900Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:14:00.195900Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"VL-RewardBench: A Challenging Benchmark for Vision-Language Generative Reward Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Bill Yuchen Lin, Chenxin An, Lei Li, Lingpeng Kong, Peiyi Wang, Qi Liu, Sujian Li, Tianyu Liu, Xuqing Yang, Yifan Song, Yuancheng Wei, Zhihui Xie","submitted_at":"2024-11-26T14:08:34Z","abstract_excerpt":"Vision-language generative reward models (VL-GenRMs) play a crucial role in aligning and evaluating multimodal AI systems, yet their own evaluation remains under-explored. Current assessment methods primarily rely on AI-annotated preference labels from traditional VL tasks, which can introduce biases and often fail to effectively challenge state-of-the-art models. To address these limitations, we introduce VL-RewardBench, a comprehensive benchmark spanning general multimodal queries, visual hallucination detection, and complex reasoning tasks. Through our AI-assisted annotation pipeline that c"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.17451","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/2411.17451/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":"2411.17451","created_at":"2026-07-05T11:14:00.195958+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.17451v2","created_at":"2026-07-05T11:14:00.195958+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.17451","created_at":"2026-07-05T11:14:00.195958+00:00"},{"alias_kind":"pith_short_12","alias_value":"FKH6QOCRNRX2","created_at":"2026-07-05T11:14:00.195958+00:00"},{"alias_kind":"pith_short_16","alias_value":"FKH6QOCRNRX2DDMS","created_at":"2026-07-05T11:14:00.195958+00:00"},{"alias_kind":"pith_short_8","alias_value":"FKH6QOCR","created_at":"2026-07-05T11:14:00.195958+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.32034","citing_title":"QVal: Cheaply Evaluating Dense Supervision Signals for Long-Horizon LLM Agents","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2504.12501","citing_title":"Reinforcement Learning from Human Feedback","ref_index":105,"is_internal_anchor":false},{"citing_arxiv_id":"2506.01937","citing_title":"RewardBench 2: Advancing Reward Model Evaluation","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2503.05236","citing_title":"Unified Reward Model for Multimodal Understanding and Generation","ref_index":40,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FKH6QOCRNRX2DDMSX2PGRZV4C2","json":"https://pith.science/pith/FKH6QOCRNRX2DDMSX2PGRZV4C2.json","graph_json":"https://pith.science/api/pith-number/FKH6QOCRNRX2DDMSX2PGRZV4C2/graph.json","events_json":"https://pith.science/api/pith-number/FKH6QOCRNRX2DDMSX2PGRZV4C2/events.json","paper":"https://pith.science/paper/FKH6QOCR"},"agent_actions":{"view_html":"https://pith.science/pith/FKH6QOCRNRX2DDMSX2PGRZV4C2","download_json":"https://pith.science/pith/FKH6QOCRNRX2DDMSX2PGRZV4C2.json","view_paper":"https://pith.science/paper/FKH6QOCR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.17451&json=true","fetch_graph":"https://pith.science/api/pith-number/FKH6QOCRNRX2DDMSX2PGRZV4C2/graph.json","fetch_events":"https://pith.science/api/pith-number/FKH6QOCRNRX2DDMSX2PGRZV4C2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FKH6QOCRNRX2DDMSX2PGRZV4C2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FKH6QOCRNRX2DDMSX2PGRZV4C2/action/storage_attestation","attest_author":"https://pith.science/pith/FKH6QOCRNRX2DDMSX2PGRZV4C2/action/author_attestation","sign_citation":"https://pith.science/pith/FKH6QOCRNRX2DDMSX2PGRZV4C2/action/citation_signature","submit_replication":"https://pith.science/pith/FKH6QOCRNRX2DDMSX2PGRZV4C2/action/replication_record"}},"created_at":"2026-07-05T11:14:00.195958+00:00","updated_at":"2026-07-05T11:14:00.195958+00:00"}