{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:KNVRFDKKFLCWDASK2A5R42N7Z3","short_pith_number":"pith:KNVRFDKK","schema_version":"1.0","canonical_sha256":"536b128d4a2ac561824ad03b1e69bfceead994cab92c022b22e4d6df1f9917bc","source":{"kind":"arxiv","id":"2504.06606","version":1},"attestation_state":"computed","paper":{"title":"Benchmarking Multimodal CoT Reward Model Stepwise by Visual Program","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Fei Wu, Juncheng Li, Minghe Gao, Shuang Chen, Siliang Tang, Tat-Seng Chua, Xuqi Liu, Yang Wu, Yueting Zhuang, Zhongqi Yue","submitted_at":"2025-04-09T06:09:40Z","abstract_excerpt":"Recent advancements in reward signal usage for Large Language Models (LLMs) are remarkable. However, significant challenges exist when transitioning reward signal to the multimodal domain, including labor-intensive annotations, over-reliance on one-step rewards, and inadequate evaluation. To address these issues, we propose SVIP, a novel approach to train a step-level multi-dimensional Chain-of-Thought~(CoT) reward model automatically. It generates code for solving visual tasks and transforms the analysis of code blocks into the evaluation of CoT step as training samples. Then, we train SVIP-R"},"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":"2504.06606","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-04-09T06:09:40Z","cross_cats_sorted":[],"title_canon_sha256":"b948ce03c011ac531c8ae74461e91eacc3cd7cd5badd86fde97547f09ecfebb1","abstract_canon_sha256":"cc748c26b260aa34e34124673a094a97c0b8e946882a90e1dd2ee275206a3bcc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:46:29.717029Z","signature_b64":"qC/DT6Oy+dK8KRYgEcclAXRdLfzyvlB8i270vnq7b0ktgQRc/ZURKdR5MfRbl1mm9bd0/bE26HjKI1l8Pad7Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"536b128d4a2ac561824ad03b1e69bfceead994cab92c022b22e4d6df1f9917bc","last_reissued_at":"2026-07-05T10:46:29.716541Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:46:29.716541Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Benchmarking Multimodal CoT Reward Model Stepwise by Visual Program","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Fei Wu, Juncheng Li, Minghe Gao, Shuang Chen, Siliang Tang, Tat-Seng Chua, Xuqi Liu, Yang Wu, Yueting Zhuang, Zhongqi Yue","submitted_at":"2025-04-09T06:09:40Z","abstract_excerpt":"Recent advancements in reward signal usage for Large Language Models (LLMs) are remarkable. However, significant challenges exist when transitioning reward signal to the multimodal domain, including labor-intensive annotations, over-reliance on one-step rewards, and inadequate evaluation. To address these issues, we propose SVIP, a novel approach to train a step-level multi-dimensional Chain-of-Thought~(CoT) reward model automatically. It generates code for solving visual tasks and transforms the analysis of code blocks into the evaluation of CoT step as training samples. Then, we train SVIP-R"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.06606","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/2504.06606/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":"2504.06606","created_at":"2026-07-05T10:46:29.716601+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.06606v1","created_at":"2026-07-05T10:46:29.716601+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.06606","created_at":"2026-07-05T10:46:29.716601+00:00"},{"alias_kind":"pith_short_12","alias_value":"KNVRFDKKFLCW","created_at":"2026-07-05T10:46:29.716601+00:00"},{"alias_kind":"pith_short_16","alias_value":"KNVRFDKKFLCWDASK","created_at":"2026-07-05T10:46:29.716601+00:00"},{"alias_kind":"pith_short_8","alias_value":"KNVRFDKK","created_at":"2026-07-05T10:46:29.716601+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2511.23253","citing_title":"AgroCoT: A Chain-of-Thought Benchmark for Evaluating Reasoning in Vision-Language Models for Agriculture","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2604.13029","citing_title":"Visual Preference Optimization with Rubric Rewards","ref_index":43,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KNVRFDKKFLCWDASK2A5R42N7Z3","json":"https://pith.science/pith/KNVRFDKKFLCWDASK2A5R42N7Z3.json","graph_json":"https://pith.science/api/pith-number/KNVRFDKKFLCWDASK2A5R42N7Z3/graph.json","events_json":"https://pith.science/api/pith-number/KNVRFDKKFLCWDASK2A5R42N7Z3/events.json","paper":"https://pith.science/paper/KNVRFDKK"},"agent_actions":{"view_html":"https://pith.science/pith/KNVRFDKKFLCWDASK2A5R42N7Z3","download_json":"https://pith.science/pith/KNVRFDKKFLCWDASK2A5R42N7Z3.json","view_paper":"https://pith.science/paper/KNVRFDKK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.06606&json=true","fetch_graph":"https://pith.science/api/pith-number/KNVRFDKKFLCWDASK2A5R42N7Z3/graph.json","fetch_events":"https://pith.science/api/pith-number/KNVRFDKKFLCWDASK2A5R42N7Z3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KNVRFDKKFLCWDASK2A5R42N7Z3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KNVRFDKKFLCWDASK2A5R42N7Z3/action/storage_attestation","attest_author":"https://pith.science/pith/KNVRFDKKFLCWDASK2A5R42N7Z3/action/author_attestation","sign_citation":"https://pith.science/pith/KNVRFDKKFLCWDASK2A5R42N7Z3/action/citation_signature","submit_replication":"https://pith.science/pith/KNVRFDKKFLCWDASK2A5R42N7Z3/action/replication_record"}},"created_at":"2026-07-05T10:46:29.716601+00:00","updated_at":"2026-07-05T10:46:29.716601+00:00"}