{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2IE2ALFJF4TDNRSCHRPG3IYJUU","short_pith_number":"pith:2IE2ALFJ","schema_version":"1.0","canonical_sha256":"d209a02ca92f2636c6423c5e6da309a520dea4de66f9453405633389f0762559","source":{"kind":"arxiv","id":"2506.18203","version":3},"attestation_state":"computed","paper":{"title":"Shrinking the Generation-Verification Gap with Weak Verifiers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Azalia Mirhoseini, Ben Athiwaratkun, Brendan McLaughlin, Christopher R\\'e, E. Kelly Buchanan, Frederic Sala, Jon Saad-Falcon, Mayee F. Chen, Scott Linderman, Shang Zhu, Tanvir Bhathal, Tzu-Heng Huang","submitted_at":"2025-06-22T23:38:15Z","abstract_excerpt":"Verifiers can improve language model capabilities by scoring and ranking responses from generated candidates. Currently, high-quality verifiers are either unscalable (e.g., humans) or limited in utility (e.g., tools like Lean). While LM judges and reward models have become broadly useful as general-purpose verifiers, a significant performance gap remains between them and oracle verifiers (verifiers with perfect accuracy). To help close this gap, we introduce Weaver, a framework for designing a strong verifier by combining multiple weak, imperfect verifiers. We find weighted ensembles of verifi"},"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":"2506.18203","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-06-22T23:38:15Z","cross_cats_sorted":[],"title_canon_sha256":"ac9bf8e947c75189bdc4d83ecd0bd8e072a6c357a434c098403414e44a6c670b","abstract_canon_sha256":"30b1972afb828116b1c7bf6246a283a79699603743c103b7e7aa7a030b920dec"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-07T00:44:58.955792Z","signature_b64":"qdx7nzJ4IRoVog+/Hw2VuCxAdpPkmIQP4pJsaWCToUPeBglB2v3H64L+vQAnNPm36oNLj58b8IUOKLyl2F8lBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d209a02ca92f2636c6423c5e6da309a520dea4de66f9453405633389f0762559","last_reissued_at":"2026-08-07T00:44:58.954271Z","signature_status":"signed_v1","first_computed_at":"2026-08-07T00:44:58.954271Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Shrinking the Generation-Verification Gap with Weak Verifiers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Azalia Mirhoseini, Ben Athiwaratkun, Brendan McLaughlin, Christopher R\\'e, E. Kelly Buchanan, Frederic Sala, Jon Saad-Falcon, Mayee F. Chen, Scott Linderman, Shang Zhu, Tanvir Bhathal, Tzu-Heng Huang","submitted_at":"2025-06-22T23:38:15Z","abstract_excerpt":"Verifiers can improve language model capabilities by scoring and ranking responses from generated candidates. Currently, high-quality verifiers are either unscalable (e.g., humans) or limited in utility (e.g., tools like Lean). While LM judges and reward models have become broadly useful as general-purpose verifiers, a significant performance gap remains between them and oracle verifiers (verifiers with perfect accuracy). To help close this gap, we introduce Weaver, a framework for designing a strong verifier by combining multiple weak, imperfect verifiers. We find weighted ensembles of verifi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.18203","kind":"arxiv","version":3},"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/2506.18203/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":"2506.18203","created_at":"2026-08-07T00:44:58.955525+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.18203v3","created_at":"2026-08-07T00:44:58.955525+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.18203","created_at":"2026-08-07T00:44:58.955525+00:00"},{"alias_kind":"pith_short_12","alias_value":"2IE2ALFJF4TD","created_at":"2026-08-07T00:44:58.955525+00:00"},{"alias_kind":"pith_short_16","alias_value":"2IE2ALFJF4TDNRSC","created_at":"2026-08-07T00:44:58.955525+00:00"},{"alias_kind":"pith_short_8","alias_value":"2IE2ALFJ","created_at":"2026-08-07T00:44:58.955525+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":6,"sample":[{"citing_arxiv_id":"2607.05391","citing_title":"LLM-as-a-Verifier: A General-Purpose Verification Framework","ref_index":62,"is_internal_anchor":true},{"citing_arxiv_id":"2606.01075","citing_title":"On the Generalization Gap in Self-Evolving Language Model Reasoning","ref_index":28,"is_internal_anchor":true},{"citing_arxiv_id":"2606.05268","citing_title":"Aggregating LLM-Based Weak Verifiers for Spatial Layout Generation","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"2509.17995","citing_title":"Variation in Verification: Understanding Verification Dynamics in Large Language Models","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"2604.18547","citing_title":"FUSE: Ensembling Verifiers with Zero Labeled Data","ref_index":9,"is_internal_anchor":true},{"citing_arxiv_id":"2604.18936","citing_title":"Fine-Tuning Small Reasoning Models for Quantum Field Theory","ref_index":59,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2IE2ALFJF4TDNRSCHRPG3IYJUU","json":"https://pith.science/pith/2IE2ALFJF4TDNRSCHRPG3IYJUU.json","graph_json":"https://pith.science/api/pith-number/2IE2ALFJF4TDNRSCHRPG3IYJUU/graph.json","events_json":"https://pith.science/api/pith-number/2IE2ALFJF4TDNRSCHRPG3IYJUU/events.json","paper":"https://pith.science/paper/2IE2ALFJ"},"agent_actions":{"view_html":"https://pith.science/pith/2IE2ALFJF4TDNRSCHRPG3IYJUU","download_json":"https://pith.science/pith/2IE2ALFJF4TDNRSCHRPG3IYJUU.json","view_paper":"https://pith.science/paper/2IE2ALFJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.18203&json=true","fetch_graph":"https://pith.science/api/pith-number/2IE2ALFJF4TDNRSCHRPG3IYJUU/graph.json","fetch_events":"https://pith.science/api/pith-number/2IE2ALFJF4TDNRSCHRPG3IYJUU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2IE2ALFJF4TDNRSCHRPG3IYJUU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2IE2ALFJF4TDNRSCHRPG3IYJUU/action/storage_attestation","attest_author":"https://pith.science/pith/2IE2ALFJF4TDNRSCHRPG3IYJUU/action/author_attestation","sign_citation":"https://pith.science/pith/2IE2ALFJF4TDNRSCHRPG3IYJUU/action/citation_signature","submit_replication":"https://pith.science/pith/2IE2ALFJF4TDNRSCHRPG3IYJUU/action/replication_record"}},"created_at":"2026-08-07T00:44:58.955525+00:00","updated_at":"2026-08-07T00:44:58.955525+00:00"}