{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:62LJAXOLTT664PUD6W3TSQRO6T","short_pith_number":"pith:62LJAXOL","schema_version":"1.0","canonical_sha256":"f696905dcb9cfdee3e83f5b739422ef4efa0434684983a9ea8ee432b258885c5","source":{"kind":"arxiv","id":"2408.03350","version":3},"attestation_state":"computed","paper":{"title":"miniCTX: Neural Theorem Proving with (Long-)Contexts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Jiewen Hu, Sean Welleck, Thomas Zhu","submitted_at":"2024-08-05T20:19:18Z","abstract_excerpt":"Real-world formal theorem proving often depends on a wealth of context, including definitions, lemmas, comments, file structure, and other information. We introduce miniCTX, which tests a model's ability to prove formal mathematical theorems that depend on new context that is not seen during training. miniCTX contains theorems sourced from real Lean projects and textbooks, each associated with a context that can span tens of thousands of tokens. Models are tasked with proving a theorem given access to code from the theorem's repository, which contains context that is needed for the proof. As a"},"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":"2408.03350","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2024-08-05T20:19:18Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"a243c419890098f3c1257f4d78f6aed75a6b5c601c20e663e639b678549e0ff9","abstract_canon_sha256":"a04b74ebfd8773f1dd044d11beffd1be854e9e79de94285603ee1e3e1f877d48"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:23:22.471988Z","signature_b64":"6QF/JmupQqWJ7i0bfzeQ1Bv7Ej+jA56WLrGM7vepPz2c1Iya1aQZ5K/9E/+AW7PYGy2oWZduB/Mwd8JxGkgoCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f696905dcb9cfdee3e83f5b739422ef4efa0434684983a9ea8ee432b258885c5","last_reissued_at":"2026-07-05T10:23:22.471054Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:23:22.471054Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"miniCTX: Neural Theorem Proving with (Long-)Contexts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Jiewen Hu, Sean Welleck, Thomas Zhu","submitted_at":"2024-08-05T20:19:18Z","abstract_excerpt":"Real-world formal theorem proving often depends on a wealth of context, including definitions, lemmas, comments, file structure, and other information. We introduce miniCTX, which tests a model's ability to prove formal mathematical theorems that depend on new context that is not seen during training. miniCTX contains theorems sourced from real Lean projects and textbooks, each associated with a context that can span tens of thousands of tokens. Models are tasked with proving a theorem given access to code from the theorem's repository, which contains context that is needed for the proof. As a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.03350","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/2408.03350/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":"2408.03350","created_at":"2026-07-05T10:23:22.471157+00:00"},{"alias_kind":"arxiv_version","alias_value":"2408.03350v3","created_at":"2026-07-05T10:23:22.471157+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.03350","created_at":"2026-07-05T10:23:22.471157+00:00"},{"alias_kind":"pith_short_12","alias_value":"62LJAXOLTT66","created_at":"2026-07-05T10:23:22.471157+00:00"},{"alias_kind":"pith_short_16","alias_value":"62LJAXOLTT664PUD","created_at":"2026-07-05T10:23:22.471157+00:00"},{"alias_kind":"pith_short_8","alias_value":"62LJAXOL","created_at":"2026-07-05T10:23:22.471157+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.01008","citing_title":"FVSpec: Real-World Property-Based Tests as Lean Challenges","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2410.04753","citing_title":"ImProver: Agent-Based Automated Proof Optimization","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2603.14628","citing_title":"s2n-bignum-bench: A practical benchmark for evaluating low-level code reasoning of LLMs","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2502.05171","citing_title":"Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach","ref_index":74,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06882","citing_title":"How Well Do LLMs Perform on the Simplest Long-Chain Reasoning Tasks: An Empirical Study on the Equivalence Class Problem","ref_index":37,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/62LJAXOLTT664PUD6W3TSQRO6T","json":"https://pith.science/pith/62LJAXOLTT664PUD6W3TSQRO6T.json","graph_json":"https://pith.science/api/pith-number/62LJAXOLTT664PUD6W3TSQRO6T/graph.json","events_json":"https://pith.science/api/pith-number/62LJAXOLTT664PUD6W3TSQRO6T/events.json","paper":"https://pith.science/paper/62LJAXOL"},"agent_actions":{"view_html":"https://pith.science/pith/62LJAXOLTT664PUD6W3TSQRO6T","download_json":"https://pith.science/pith/62LJAXOLTT664PUD6W3TSQRO6T.json","view_paper":"https://pith.science/paper/62LJAXOL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2408.03350&json=true","fetch_graph":"https://pith.science/api/pith-number/62LJAXOLTT664PUD6W3TSQRO6T/graph.json","fetch_events":"https://pith.science/api/pith-number/62LJAXOLTT664PUD6W3TSQRO6T/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/62LJAXOLTT664PUD6W3TSQRO6T/action/timestamp_anchor","attest_storage":"https://pith.science/pith/62LJAXOLTT664PUD6W3TSQRO6T/action/storage_attestation","attest_author":"https://pith.science/pith/62LJAXOLTT664PUD6W3TSQRO6T/action/author_attestation","sign_citation":"https://pith.science/pith/62LJAXOLTT664PUD6W3TSQRO6T/action/citation_signature","submit_replication":"https://pith.science/pith/62LJAXOLTT664PUD6W3TSQRO6T/action/replication_record"}},"created_at":"2026-07-05T10:23:22.471157+00:00","updated_at":"2026-07-05T10:23:22.471157+00:00"}