{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:FF3TW7VFG65H2CGE4OLCTH36OG","short_pith_number":"pith:FF3TW7VF","schema_version":"1.0","canonical_sha256":"29773b7ea537ba7d08c4e396299f7e71b1ed4234a2b6df132a0e0f37deccb6f0","source":{"kind":"arxiv","id":"2502.00271","version":1},"attestation_state":"computed","paper":{"title":"Scaling Flaws of Verifier-Guided Search in Mathematical Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Benyou Wang, Fei Yu, Yingru Li","submitted_at":"2025-02-01T02:08:49Z","abstract_excerpt":"Large language models (LLMs) struggle with multi-step reasoning, where inference-time scaling has emerged as a promising strategy for performance improvement. Verifier-guided search outperforms repeated sampling when sample size is limited by selecting and prioritizing valid reasoning paths. However, we identify a critical limitation: scaling flaws, prevalent across different models (Mistral 7B and DeepSeekMath 7B), benchmarks (GSM8K and MATH), and verifiers (outcome value models and process reward models). As sample size increases, verifier-guided search exhibits diminishing advantages and ev"},"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":"2502.00271","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-02-01T02:08:49Z","cross_cats_sorted":[],"title_canon_sha256":"6e34e67c1862a534075985183015b5bcea4bef8d63ce5077e39bdae2770edec9","abstract_canon_sha256":"a0234d70af24b2af5bfc9a535afd1e07af399498c9c09040316c3527126185b1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:08:31.516789Z","signature_b64":"KSut3WQ4D0+6/9wbbBz7PwX45Ahz6NCO2LGiGIdVlcNaE7DH77wN41j3GksOV3TZU9LMgAeJ2fHy9At33qMVDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"29773b7ea537ba7d08c4e396299f7e71b1ed4234a2b6df132a0e0f37deccb6f0","last_reissued_at":"2026-07-05T10:08:31.516379Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:08:31.516379Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Scaling Flaws of Verifier-Guided Search in Mathematical Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Benyou Wang, Fei Yu, Yingru Li","submitted_at":"2025-02-01T02:08:49Z","abstract_excerpt":"Large language models (LLMs) struggle with multi-step reasoning, where inference-time scaling has emerged as a promising strategy for performance improvement. Verifier-guided search outperforms repeated sampling when sample size is limited by selecting and prioritizing valid reasoning paths. However, we identify a critical limitation: scaling flaws, prevalent across different models (Mistral 7B and DeepSeekMath 7B), benchmarks (GSM8K and MATH), and verifiers (outcome value models and process reward models). As sample size increases, verifier-guided search exhibits diminishing advantages and ev"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.00271","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/2502.00271/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":"2502.00271","created_at":"2026-07-05T10:08:31.516436+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.00271v1","created_at":"2026-07-05T10:08:31.516436+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.00271","created_at":"2026-07-05T10:08:31.516436+00:00"},{"alias_kind":"pith_short_12","alias_value":"FF3TW7VFG65H","created_at":"2026-07-05T10:08:31.516436+00:00"},{"alias_kind":"pith_short_16","alias_value":"FF3TW7VFG65H2CGE","created_at":"2026-07-05T10:08:31.516436+00:00"},{"alias_kind":"pith_short_8","alias_value":"FF3TW7VF","created_at":"2026-07-05T10:08:31.516436+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.08256","citing_title":"Best-of-$N$ TTS Evaluation is Confounded by ASR Family Alignment","ref_index":11,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FF3TW7VFG65H2CGE4OLCTH36OG","json":"https://pith.science/pith/FF3TW7VFG65H2CGE4OLCTH36OG.json","graph_json":"https://pith.science/api/pith-number/FF3TW7VFG65H2CGE4OLCTH36OG/graph.json","events_json":"https://pith.science/api/pith-number/FF3TW7VFG65H2CGE4OLCTH36OG/events.json","paper":"https://pith.science/paper/FF3TW7VF"},"agent_actions":{"view_html":"https://pith.science/pith/FF3TW7VFG65H2CGE4OLCTH36OG","download_json":"https://pith.science/pith/FF3TW7VFG65H2CGE4OLCTH36OG.json","view_paper":"https://pith.science/paper/FF3TW7VF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.00271&json=true","fetch_graph":"https://pith.science/api/pith-number/FF3TW7VFG65H2CGE4OLCTH36OG/graph.json","fetch_events":"https://pith.science/api/pith-number/FF3TW7VFG65H2CGE4OLCTH36OG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FF3TW7VFG65H2CGE4OLCTH36OG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FF3TW7VFG65H2CGE4OLCTH36OG/action/storage_attestation","attest_author":"https://pith.science/pith/FF3TW7VFG65H2CGE4OLCTH36OG/action/author_attestation","sign_citation":"https://pith.science/pith/FF3TW7VFG65H2CGE4OLCTH36OG/action/citation_signature","submit_replication":"https://pith.science/pith/FF3TW7VFG65H2CGE4OLCTH36OG/action/replication_record"}},"created_at":"2026-07-05T10:08:31.516436+00:00","updated_at":"2026-07-05T10:08:31.516436+00:00"}