{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:4JZOVUM5DEXOJPFR6BJCVXPLGK","short_pith_number":"pith:4JZOVUM5","schema_version":"1.0","canonical_sha256":"e272ead19d192ee4bcb1f0522addeb3293e4427ec8adb44786f12e3d3251cd23","source":{"kind":"arxiv","id":"2109.00110","version":2},"attestation_state":"computed","paper":{"title":"MiniF2F: a cross-system benchmark for formal Olympiad-level mathematics","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"MiniF2F supplies 488 formal statements from math contests as a shared benchmark for neural theorem provers across systems.","cross_cats":["cs.FL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Jesse Michael Han, Kunhao Zheng, Stanislas Polu","submitted_at":"2021-08-31T23:21:12Z","abstract_excerpt":"We present miniF2F, a dataset of formal Olympiad-level mathematics problems statements intended to provide a unified cross-system benchmark for neural theorem proving. The miniF2F benchmark currently targets Metamath, Lean, Isabelle (partially) and HOL Light (partially) and consists of 488 problem statements drawn from the AIME, AMC, and the International Mathematical Olympiad (IMO), as well as material from high-school and undergraduate mathematics courses. We report baseline results using GPT-f, a neural theorem prover based on GPT-3 and provide an analysis of its performance. We intend for "},"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":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2109.00110","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2021-08-31T23:21:12Z","cross_cats_sorted":["cs.FL","cs.LG"],"title_canon_sha256":"cc0244c5959f2c66a5c6d38b722450654424231e35e121c472593706bd580a10","abstract_canon_sha256":"1135ccff970ede9810711c2e659e388f6309e7db620a343c59e92ac01865a508"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:46.754758Z","signature_b64":"wAW6p6JnFXDT/VKFV75Rs9tCX8n1oVD3I/VhZ+0ayAdEP5IJmN+SygaIpPiJf8wHP28dZ6dSy4gTIPV/6o3dDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e272ead19d192ee4bcb1f0522addeb3293e4427ec8adb44786f12e3d3251cd23","last_reissued_at":"2026-05-17T23:38:46.754275Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:46.754275Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MiniF2F: a cross-system benchmark for formal Olympiad-level mathematics","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"MiniF2F supplies 488 formal statements from math contests as a shared benchmark for neural theorem provers across systems.","cross_cats":["cs.FL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Jesse Michael Han, Kunhao Zheng, Stanislas Polu","submitted_at":"2021-08-31T23:21:12Z","abstract_excerpt":"We present miniF2F, a dataset of formal Olympiad-level mathematics problems statements intended to provide a unified cross-system benchmark for neural theorem proving. The miniF2F benchmark currently targets Metamath, Lean, Isabelle (partially) and HOL Light (partially) and consists of 488 problem statements drawn from the AIME, AMC, and the International Mathematical Olympiad (IMO), as well as material from high-school and undergraduate mathematics courses. We report baseline results using GPT-f, a neural theorem prover based on GPT-3 and provide an analysis of its performance. We intend for "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We present miniF2F, a dataset of formal Olympiad-level mathematics problems statements intended to provide a unified cross-system benchmark for neural theorem proving.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the chosen contest problems can be accurately translated into correct formal statements in each target system and that these statements form a representative and stable measure of progress for neural theorem provers.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"MiniF2F is a new cross-system benchmark containing 488 Olympiad-level mathematics problems formalized in Metamath, Lean, Isabelle, and HOL Light, together with baseline results from a GPT-3-based prover.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"MiniF2F supplies 488 formal statements from math contests as a shared benchmark for neural theorem provers across systems.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"c87da5ba191ff44df33fb435f95a5f9ddd18aca31db22e3705fa52217b583363"},"source":{"id":"2109.00110","kind":"arxiv","version":2},"verdict":{"id":"8575ac97-35f6-45e5-b41d-e6e18ca0cfbd","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T19:59:49.058006Z","strongest_claim":"We present miniF2F, a dataset of formal Olympiad-level mathematics problems statements intended to provide a unified cross-system benchmark for neural theorem proving.","one_line_summary":"MiniF2F is a new cross-system benchmark containing 488 Olympiad-level mathematics problems formalized in Metamath, Lean, Isabelle, and HOL Light, together with baseline results from a GPT-3-based prover.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the chosen contest problems can be accurately translated into correct formal statements in each target system and that these statements form a representative and stable measure of progress for neural theorem provers.","pith_extraction_headline":"MiniF2F supplies 488 formal statements from math contests as a shared benchmark for neural theorem provers across systems."},"references":{"count":18,"sample":[{"doi":"","year":1905,"title":"Holist: An envi- ronment for machine learning of higher order logic theorem proving","work_id":"f5bfea42-9a09-44a0-a9c5-6f1c4ac3ae2a","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"Kevin Buzzard, Johan Commelin, and Patrick Massot","work_id":"39b021ec-2550-4d2e-a7b8-256e275589a2","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2009,"title":"Imagenet: A large-scale hi- erarchical image database","work_id":"07e79ded-6939-4deb-a631-4cb7a12164b4","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Proof artifact co-training for theorem proving with language models.arXiv preprint arXiv:2102.06203","work_id":"86d8222e-af2a-450c-b058-95a4de0be853","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Measuring Mathematical Problem Solving With the MATH Dataset","work_id":"50652ac6-fb7c-4675-a2c2-159c241feb17","ref_index":5,"cited_arxiv_id":"2103.03874","is_internal_anchor":true}],"resolved_work":18,"snapshot_sha256":"736bfea96116f93232f41e4dadd248a1c8872bf77d05991e093fc5fcac70debb","internal_anchors":2},"formal_canon":{"evidence_count":1,"snapshot_sha256":"fad9ba61dfc0242697864df25c2ee1455fe66f1053de48c0d9a43ed60422dfcc"},"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":"2109.00110","created_at":"2026-05-17T23:38:46.754356+00:00"},{"alias_kind":"arxiv_version","alias_value":"2109.00110v2","created_at":"2026-05-17T23:38:46.754356+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2109.00110","created_at":"2026-05-17T23:38:46.754356+00:00"},{"alias_kind":"pith_short_12","alias_value":"4JZOVUM5DEXO","created_at":"2026-05-18T12:33:33.725879+00:00"},{"alias_kind":"pith_short_16","alias_value":"4JZOVUM5DEXOJPFR","created_at":"2026-05-18T12:33:33.725879+00:00"},{"alias_kind":"pith_short_8","alias_value":"4JZOVUM5","created_at":"2026-05-18T12:33:33.725879+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":24,"internal_anchor_count":24,"sample":[{"citing_arxiv_id":"2510.12787","citing_title":"Ax-Prover: A Deep Reasoning Agentic Framework for Theorem Proving in Mathematics and Quantum Physics","ref_index":73,"is_internal_anchor":true},{"citing_arxiv_id":"2402.03300","citing_title":"DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models","ref_index":60,"is_internal_anchor":true},{"citing_arxiv_id":"2605.17283","citing_title":"OProver: A Unified Framework for Agentic Formal Theorem Proving","ref_index":124,"is_internal_anchor":true},{"citing_arxiv_id":"2605.17255","citing_title":"CAM-Bench: A Benchmark for Computational and Applied Mathematics in Lean","ref_index":43,"is_internal_anchor":true},{"citing_arxiv_id":"2605.20120","citing_title":"Using Aristotle API for AI-Assisted Theorem Proving in Lean 4: A Formalisation Case Study of the Grasshopper Problem","ref_index":12,"is_internal_anchor":true},{"citing_arxiv_id":"2310.10631","citing_title":"Llemma: An Open Language Model For Mathematics","ref_index":203,"is_internal_anchor":true},{"citing_arxiv_id":"2601.13209","citing_title":"AI for Mathematics: Progress, Challenges, and Prospects","ref_index":175,"is_internal_anchor":true},{"citing_arxiv_id":"2602.24273","citing_title":"A Minimal Agent for Automated Theorem Proving","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"2410.07985","citing_title":"Omni-MATH: A Universal Olympiad Level Mathematic Benchmark For Large Language Models","ref_index":79,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14061","citing_title":"MathAtlas: A Benchmark for Autoformalization in the Wild","ref_index":42,"is_internal_anchor":true},{"citing_arxiv_id":"2605.13171","citing_title":"Formal Conjectures: An Open and Evolving Benchmark for Verified Discovery in Mathematics","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"2604.02709","citing_title":"Evaluating the Formal Reasoning Capabilities of Large Language Models through Chomsky Hierarchy","ref_index":59,"is_internal_anchor":true},{"citing_arxiv_id":"2605.11905","citing_title":"Rethinking Supervision Granularity: Segment-Level Learning for LLM-Based Theorem Proving","ref_index":12,"is_internal_anchor":true},{"citing_arxiv_id":"2604.25031","citing_title":"Faithful Autoformalization via Roundtrip Verification and Repair","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2605.09292","citing_title":"Beyond Accuracy: Evaluating Strategy Diversity in LLM Mathematical Reasoning","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2407.21787","citing_title":"Large Language Monkeys: Scaling Inference Compute with Repeated Sampling","ref_index":68,"is_internal_anchor":true},{"citing_arxiv_id":"2605.09012","citing_title":"Re$^2$Math: Benchmarking Theorem Retrieval in Research-Level Mathematics","ref_index":26,"is_internal_anchor":true},{"citing_arxiv_id":"2605.08498","citing_title":"MathConstraint: Automated Generation of Verified Combinatorial Reasoning Instances for LLMs","ref_index":66,"is_internal_anchor":true},{"citing_arxiv_id":"2604.25155","citing_title":"Rethinking Wireless Communications through Formal Mathematical AI Reasoning","ref_index":26,"is_internal_anchor":true},{"citing_arxiv_id":"2604.25031","citing_title":"Faithful Autoformalization via Roundtrip Verification and Repair","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2604.23712","citing_title":"OptProver: Bridging Olympiad and Optimization through Continual Training in Formal Theorem Proving","ref_index":35,"is_internal_anchor":true},{"citing_arxiv_id":"2604.20622","citing_title":"pAI/MSc: ML Theory Research with Humans on the Loop","ref_index":85,"is_internal_anchor":true},{"citing_arxiv_id":"2604.06401","citing_title":"ProofSketcher: Hybrid LLM + Lightweight Proof Checker for Reliable Math/Logic Reasoning","ref_index":26,"is_internal_anchor":true},{"citing_arxiv_id":"2604.16278","citing_title":"Learning to Reason with Insight for Informal Theorem Proving","ref_index":3,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":1,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4JZOVUM5DEXOJPFR6BJCVXPLGK","json":"https://pith.science/pith/4JZOVUM5DEXOJPFR6BJCVXPLGK.json","graph_json":"https://pith.science/api/pith-number/4JZOVUM5DEXOJPFR6BJCVXPLGK/graph.json","events_json":"https://pith.science/api/pith-number/4JZOVUM5DEXOJPFR6BJCVXPLGK/events.json","paper":"https://pith.science/paper/4JZOVUM5"},"agent_actions":{"view_html":"https://pith.science/pith/4JZOVUM5DEXOJPFR6BJCVXPLGK","download_json":"https://pith.science/pith/4JZOVUM5DEXOJPFR6BJCVXPLGK.json","view_paper":"https://pith.science/paper/4JZOVUM5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2109.00110&json=true","fetch_graph":"https://pith.science/api/pith-number/4JZOVUM5DEXOJPFR6BJCVXPLGK/graph.json","fetch_events":"https://pith.science/api/pith-number/4JZOVUM5DEXOJPFR6BJCVXPLGK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4JZOVUM5DEXOJPFR6BJCVXPLGK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4JZOVUM5DEXOJPFR6BJCVXPLGK/action/storage_attestation","attest_author":"https://pith.science/pith/4JZOVUM5DEXOJPFR6BJCVXPLGK/action/author_attestation","sign_citation":"https://pith.science/pith/4JZOVUM5DEXOJPFR6BJCVXPLGK/action/citation_signature","submit_replication":"https://pith.science/pith/4JZOVUM5DEXOJPFR6BJCVXPLGK/action/replication_record"}},"created_at":"2026-05-17T23:38:46.754356+00:00","updated_at":"2026-05-17T23:38:46.754356+00:00"}