{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:CA7RJKZUMYY6C2TNTAKCBFNKWE","short_pith_number":"pith:CA7RJKZU","schema_version":"1.0","canonical_sha256":"103f14ab346631e16a6d98142095aab10af4fcdf67a76729e61b4227dfae7a0e","source":{"kind":"arxiv","id":"2506.13888","version":1},"attestation_state":"computed","paper":{"title":"VL-GenRM: Enhancing Vision-Language Verification via Vision Experts and Iterative Training","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.CL","authors_text":"Jipeng Zhang, Kehao Miao, Renjie Pi, Rui Pan, Runtao Liu, Tong Zhang, Zhaowei Wang","submitted_at":"2025-06-16T18:10:51Z","abstract_excerpt":"Reinforcement Fine-Tuning (RFT) with verifiable rewards has advanced large language models but remains underexplored for Vision-Language (VL) models. The Vision-Language Reward Model (VL-RM) is key to aligning VL models by providing structured feedback, yet training effective VL-RMs faces two major challenges. First, the bootstrapping dilemma arises as high-quality training data depends on already strong VL models, creating a cycle where self-generated supervision reinforces existing biases. Second, modality bias and negative example amplification occur when VL models hallucinate incorrect vis"},"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":"2506.13888","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2025-06-16T18:10:51Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"58a25075b40bc9dcfbeef9298bfed80b7703b1dba3bba4d7b00fc2c6fab80337","abstract_canon_sha256":"d659400435053a97a3d1fdcd1649b24e36800ea574e2faaa849e15c56e45a7dd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:22:45.251942Z","signature_b64":"mbT+E333S3yUMoaNndpsznsNNNTQCHH/bMhSq9l213Jh4EK+gQ4844R0m5dU8rss7X72sR5JV/Z360j/vtNNAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"103f14ab346631e16a6d98142095aab10af4fcdf67a76729e61b4227dfae7a0e","last_reissued_at":"2026-07-05T11:22:45.251481Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:22:45.251481Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"VL-GenRM: Enhancing Vision-Language Verification via Vision Experts and Iterative Training","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.CL","authors_text":"Jipeng Zhang, Kehao Miao, Renjie Pi, Rui Pan, Runtao Liu, Tong Zhang, Zhaowei Wang","submitted_at":"2025-06-16T18:10:51Z","abstract_excerpt":"Reinforcement Fine-Tuning (RFT) with verifiable rewards has advanced large language models but remains underexplored for Vision-Language (VL) models. The Vision-Language Reward Model (VL-RM) is key to aligning VL models by providing structured feedback, yet training effective VL-RMs faces two major challenges. First, the bootstrapping dilemma arises as high-quality training data depends on already strong VL models, creating a cycle where self-generated supervision reinforces existing biases. Second, modality bias and negative example amplification occur when VL models hallucinate incorrect vis"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.13888","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/2506.13888/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":"2506.13888","created_at":"2026-07-05T11:22:45.251533+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.13888v1","created_at":"2026-07-05T11:22:45.251533+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.13888","created_at":"2026-07-05T11:22:45.251533+00:00"},{"alias_kind":"pith_short_12","alias_value":"CA7RJKZUMYY6","created_at":"2026-07-05T11:22:45.251533+00:00"},{"alias_kind":"pith_short_16","alias_value":"CA7RJKZUMYY6C2TN","created_at":"2026-07-05T11:22:45.251533+00:00"},{"alias_kind":"pith_short_8","alias_value":"CA7RJKZU","created_at":"2026-07-05T11:22:45.251533+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.24583","citing_title":"Improving Vision-language Models with Perception-centric Process Reward Models","ref_index":48,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CA7RJKZUMYY6C2TNTAKCBFNKWE","json":"https://pith.science/pith/CA7RJKZUMYY6C2TNTAKCBFNKWE.json","graph_json":"https://pith.science/api/pith-number/CA7RJKZUMYY6C2TNTAKCBFNKWE/graph.json","events_json":"https://pith.science/api/pith-number/CA7RJKZUMYY6C2TNTAKCBFNKWE/events.json","paper":"https://pith.science/paper/CA7RJKZU"},"agent_actions":{"view_html":"https://pith.science/pith/CA7RJKZUMYY6C2TNTAKCBFNKWE","download_json":"https://pith.science/pith/CA7RJKZUMYY6C2TNTAKCBFNKWE.json","view_paper":"https://pith.science/paper/CA7RJKZU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.13888&json=true","fetch_graph":"https://pith.science/api/pith-number/CA7RJKZUMYY6C2TNTAKCBFNKWE/graph.json","fetch_events":"https://pith.science/api/pith-number/CA7RJKZUMYY6C2TNTAKCBFNKWE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CA7RJKZUMYY6C2TNTAKCBFNKWE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CA7RJKZUMYY6C2TNTAKCBFNKWE/action/storage_attestation","attest_author":"https://pith.science/pith/CA7RJKZUMYY6C2TNTAKCBFNKWE/action/author_attestation","sign_citation":"https://pith.science/pith/CA7RJKZUMYY6C2TNTAKCBFNKWE/action/citation_signature","submit_replication":"https://pith.science/pith/CA7RJKZUMYY6C2TNTAKCBFNKWE/action/replication_record"}},"created_at":"2026-07-05T11:22:45.251533+00:00","updated_at":"2026-07-05T11:22:45.251533+00:00"}