{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:7ZCVVK5XHE3Q5CZAGYLDRIIRCJ","short_pith_number":"pith:7ZCVVK5X","schema_version":"1.0","canonical_sha256":"fe455aabb739370e8b20361638a11112422f559d46fd0c1f2fee9c8a14e27ea3","source":{"kind":"arxiv","id":"2504.10222","version":1},"attestation_state":"computed","paper":{"title":"PRM-BAS: Enhancing Multimodal Reasoning through PRM-guided Beam Annealing Search","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.MM","authors_text":"Feng Ma, Jianqing Gao, Jianshu Zhang, Jiefeng Ma, Jun Du, Pengfei Hu, Qikai Chang, Qingfeng Liu, Quan Liu, Shuhang Liu, Zhenrong Zhang","submitted_at":"2025-04-14T13:44:11Z","abstract_excerpt":"Recent work increasingly focuses on improving the reasoning capabilities of Multimodal Large Language Models (MLLMs). Among existing methods, Process Reward Models (PRMs) stand out for offering dense, step-wise supervision to guide intermediate reasoning. However, how to effectively integrate PRMs into search strategies remains an open question. In this paper, we introduce PRM-BAS (PRM-Guided Beam Annealing Search), a lightweight approach for PRM-guided reasoning that dynamically adjusts beam size -- starting with a broader search space and gradually narrowing it as contextual information accu"},"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":"2504.10222","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.MM","submitted_at":"2025-04-14T13:44:11Z","cross_cats_sorted":[],"title_canon_sha256":"efb512a2cdfd92c260b005a09f2387ec2f9b2eb4cff6b341a3ddcfbb16ec257c","abstract_canon_sha256":"dda0c2df68b22e214355678055e3d7bca383b14f728054c21190cc8f00b8035c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:48:49.423357Z","signature_b64":"CD15WXrjQ5LaEAAciKtZDO9NOgiwIwLX8vh4j8bhoHCNdN4sRbo5S1rmZ0RJ8DTwCfsCNCRynpahlvtzt2BYAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fe455aabb739370e8b20361638a11112422f559d46fd0c1f2fee9c8a14e27ea3","last_reissued_at":"2026-07-05T10:48:49.422953Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:48:49.422953Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PRM-BAS: Enhancing Multimodal Reasoning through PRM-guided Beam Annealing Search","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.MM","authors_text":"Feng Ma, Jianqing Gao, Jianshu Zhang, Jiefeng Ma, Jun Du, Pengfei Hu, Qikai Chang, Qingfeng Liu, Quan Liu, Shuhang Liu, Zhenrong Zhang","submitted_at":"2025-04-14T13:44:11Z","abstract_excerpt":"Recent work increasingly focuses on improving the reasoning capabilities of Multimodal Large Language Models (MLLMs). Among existing methods, Process Reward Models (PRMs) stand out for offering dense, step-wise supervision to guide intermediate reasoning. However, how to effectively integrate PRMs into search strategies remains an open question. In this paper, we introduce PRM-BAS (PRM-Guided Beam Annealing Search), a lightweight approach for PRM-guided reasoning that dynamically adjusts beam size -- starting with a broader search space and gradually narrowing it as contextual information accu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.10222","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/2504.10222/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":"2504.10222","created_at":"2026-07-05T10:48:49.423004+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.10222v1","created_at":"2026-07-05T10:48:49.423004+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.10222","created_at":"2026-07-05T10:48:49.423004+00:00"},{"alias_kind":"pith_short_12","alias_value":"7ZCVVK5XHE3Q","created_at":"2026-07-05T10:48:49.423004+00:00"},{"alias_kind":"pith_short_16","alias_value":"7ZCVVK5XHE3Q5CZA","created_at":"2026-07-05T10:48:49.423004+00:00"},{"alias_kind":"pith_short_8","alias_value":"7ZCVVK5X","created_at":"2026-07-05T10:48:49.423004+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.15529","citing_title":"Process Rewards with Learned Reliability","ref_index":21,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7ZCVVK5XHE3Q5CZAGYLDRIIRCJ","json":"https://pith.science/pith/7ZCVVK5XHE3Q5CZAGYLDRIIRCJ.json","graph_json":"https://pith.science/api/pith-number/7ZCVVK5XHE3Q5CZAGYLDRIIRCJ/graph.json","events_json":"https://pith.science/api/pith-number/7ZCVVK5XHE3Q5CZAGYLDRIIRCJ/events.json","paper":"https://pith.science/paper/7ZCVVK5X"},"agent_actions":{"view_html":"https://pith.science/pith/7ZCVVK5XHE3Q5CZAGYLDRIIRCJ","download_json":"https://pith.science/pith/7ZCVVK5XHE3Q5CZAGYLDRIIRCJ.json","view_paper":"https://pith.science/paper/7ZCVVK5X","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.10222&json=true","fetch_graph":"https://pith.science/api/pith-number/7ZCVVK5XHE3Q5CZAGYLDRIIRCJ/graph.json","fetch_events":"https://pith.science/api/pith-number/7ZCVVK5XHE3Q5CZAGYLDRIIRCJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7ZCVVK5XHE3Q5CZAGYLDRIIRCJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7ZCVVK5XHE3Q5CZAGYLDRIIRCJ/action/storage_attestation","attest_author":"https://pith.science/pith/7ZCVVK5XHE3Q5CZAGYLDRIIRCJ/action/author_attestation","sign_citation":"https://pith.science/pith/7ZCVVK5XHE3Q5CZAGYLDRIIRCJ/action/citation_signature","submit_replication":"https://pith.science/pith/7ZCVVK5XHE3Q5CZAGYLDRIIRCJ/action/replication_record"}},"created_at":"2026-07-05T10:48:49.423004+00:00","updated_at":"2026-07-05T10:48:49.423004+00:00"}