{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:6WXWBLLSDPC6RB5OOS263GCSYJ","short_pith_number":"pith:6WXWBLLS","schema_version":"1.0","canonical_sha256":"f5af60ad721bc5e887ae74b5ed9852c26d70fa4972b71131fa8eb3fb2730cf62","source":{"kind":"arxiv","id":"2507.22069","version":2},"attestation_state":"computed","paper":{"title":"A Compute-Matched Re-Evaluation of TroVE on MATH","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.PL","authors_text":"Christian Bartelt, Ian Berlot-Attwell, Janis Zenkner, Tobias Sesterhenn","submitted_at":"2025-07-16T03:11:43Z","abstract_excerpt":"Reusing established theorems and formulas is central to mathematical problem solving, serving as essential building blocks for tackling increasingly complex challenges. Recent work, TroVE, argues that code-generating Large Language Models (LLMs) can benefit similarly on the MATH benchmark by inducing and reusing higher-level toolboxes. By allocating computational budget across an ensemble of three modes -- directly generating code, creating tools, and reusing tools -- TroVE claims to outperform a PRIMITIVE baseline that only performs direct generation. However, recent analysis (Berlot-Attwell "},"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":"2507.22069","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.PL","submitted_at":"2025-07-16T03:11:43Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"479b6938cf3149b6e2d3c81ab6f2f24a52965222582ee610dc47af85f3ec49a4","abstract_canon_sha256":"fcc76777a217476db059cba368e2818a9a4878d890f4202b6232051e1e62ead3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:46:00.385055Z","signature_b64":"RmqlWc9epijwM5YIb0G/83qzlKV5Yxb+uZw1CBkTcPX6f3dXmFXWrzvU4Qd4bgtqiBNONBfbtklOyrkZi85jAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f5af60ad721bc5e887ae74b5ed9852c26d70fa4972b71131fa8eb3fb2730cf62","last_reissued_at":"2026-07-05T11:46:00.384528Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:46:00.384528Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Compute-Matched Re-Evaluation of TroVE on MATH","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.PL","authors_text":"Christian Bartelt, Ian Berlot-Attwell, Janis Zenkner, Tobias Sesterhenn","submitted_at":"2025-07-16T03:11:43Z","abstract_excerpt":"Reusing established theorems and formulas is central to mathematical problem solving, serving as essential building blocks for tackling increasingly complex challenges. Recent work, TroVE, argues that code-generating Large Language Models (LLMs) can benefit similarly on the MATH benchmark by inducing and reusing higher-level toolboxes. By allocating computational budget across an ensemble of three modes -- directly generating code, creating tools, and reusing tools -- TroVE claims to outperform a PRIMITIVE baseline that only performs direct generation. However, recent analysis (Berlot-Attwell "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.22069","kind":"arxiv","version":2},"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/2507.22069/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":"2507.22069","created_at":"2026-07-05T11:46:00.384593+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.22069v2","created_at":"2026-07-05T11:46:00.384593+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.22069","created_at":"2026-07-05T11:46:00.384593+00:00"},{"alias_kind":"pith_short_12","alias_value":"6WXWBLLSDPC6","created_at":"2026-07-05T11:46:00.384593+00:00"},{"alias_kind":"pith_short_16","alias_value":"6WXWBLLSDPC6RB5O","created_at":"2026-07-05T11:46:00.384593+00:00"},{"alias_kind":"pith_short_8","alias_value":"6WXWBLLS","created_at":"2026-07-05T11:46:00.384593+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":3,"sample":[{"citing_arxiv_id":"2603.19715","citing_title":"Neuro-Symbolic Proof Generation for Scaling Systems Software Verification","ref_index":23,"is_internal_anchor":true},{"citing_arxiv_id":"2605.05485","citing_title":"ReaComp: Compiling LLM Reasoning into Symbolic Solvers for Efficient Program Synthesis","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2605.01698","citing_title":"BIM Information Extraction Through LLM-based Adaptive Exploration","ref_index":44,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6WXWBLLSDPC6RB5OOS263GCSYJ","json":"https://pith.science/pith/6WXWBLLSDPC6RB5OOS263GCSYJ.json","graph_json":"https://pith.science/api/pith-number/6WXWBLLSDPC6RB5OOS263GCSYJ/graph.json","events_json":"https://pith.science/api/pith-number/6WXWBLLSDPC6RB5OOS263GCSYJ/events.json","paper":"https://pith.science/paper/6WXWBLLS"},"agent_actions":{"view_html":"https://pith.science/pith/6WXWBLLSDPC6RB5OOS263GCSYJ","download_json":"https://pith.science/pith/6WXWBLLSDPC6RB5OOS263GCSYJ.json","view_paper":"https://pith.science/paper/6WXWBLLS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.22069&json=true","fetch_graph":"https://pith.science/api/pith-number/6WXWBLLSDPC6RB5OOS263GCSYJ/graph.json","fetch_events":"https://pith.science/api/pith-number/6WXWBLLSDPC6RB5OOS263GCSYJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6WXWBLLSDPC6RB5OOS263GCSYJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6WXWBLLSDPC6RB5OOS263GCSYJ/action/storage_attestation","attest_author":"https://pith.science/pith/6WXWBLLSDPC6RB5OOS263GCSYJ/action/author_attestation","sign_citation":"https://pith.science/pith/6WXWBLLSDPC6RB5OOS263GCSYJ/action/citation_signature","submit_replication":"https://pith.science/pith/6WXWBLLSDPC6RB5OOS263GCSYJ/action/replication_record"}},"created_at":"2026-07-05T11:46:00.384593+00:00","updated_at":"2026-07-05T11:46:00.384593+00:00"}