{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:JYTTSXB5SVM2O5EZAR35QLLCN2","short_pith_number":"pith:JYTTSXB5","schema_version":"1.0","canonical_sha256":"4e27395c3d9559a774990477d82d626e96865e27e416af399823cded97984e3c","source":{"kind":"arxiv","id":"2502.17578","version":1},"attestation_state":"computed","paper":{"title":"How Do Large Language Monkeys Get Their Power (Laws)?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Aengus Lynch, Azalia Mirhoseini, Erik Jones, John Hughes, Jordan Juravsky, Joshua Kazdan, Robert Kirk, Rylan Schaeffer, Sanmi Koyejo, Sara Price","submitted_at":"2025-02-24T19:01:47Z","abstract_excerpt":"Recent research across mathematical problem solving, proof assistant programming and multimodal jailbreaking documents a striking finding: when (multimodal) language model tackle a suite of tasks with multiple attempts per task -- succeeding if any attempt is correct -- then the negative log of the average success rate scales a power law in the number of attempts. In this work, we identify an apparent puzzle: a simple mathematical calculation predicts that on each problem, the failure rate should fall exponentially with the number of attempts. We confirm this prediction empirically, raising 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":"2502.17578","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-02-24T19:01:47Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"9ad004f11a759ad60a2c2c42a1a548fea30bf1ca9a6f1154e5bc7f3075182698","abstract_canon_sha256":"95e37edd505cccc47033b3dab65e676c7b302b405637d5612d3e119c5ef5b6e4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:19:12.945562Z","signature_b64":"N1XrJhxmS3Vhij5qb6+cEFkBVbL2TiDLe8Zp8Vo64HA1KmWxRkKdBBlYs/H/taSR9Gg5RXEWHt16lfCF5UAgBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4e27395c3d9559a774990477d82d626e96865e27e416af399823cded97984e3c","last_reissued_at":"2026-07-05T10:19:12.945034Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:19:12.945034Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"How Do Large Language Monkeys Get Their Power (Laws)?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Aengus Lynch, Azalia Mirhoseini, Erik Jones, John Hughes, Jordan Juravsky, Joshua Kazdan, Robert Kirk, Rylan Schaeffer, Sanmi Koyejo, Sara Price","submitted_at":"2025-02-24T19:01:47Z","abstract_excerpt":"Recent research across mathematical problem solving, proof assistant programming and multimodal jailbreaking documents a striking finding: when (multimodal) language model tackle a suite of tasks with multiple attempts per task -- succeeding if any attempt is correct -- then the negative log of the average success rate scales a power law in the number of attempts. In this work, we identify an apparent puzzle: a simple mathematical calculation predicts that on each problem, the failure rate should fall exponentially with the number of attempts. We confirm this prediction empirically, raising a "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.17578","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/2502.17578/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":"2502.17578","created_at":"2026-07-05T10:19:12.945106+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.17578v1","created_at":"2026-07-05T10:19:12.945106+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.17578","created_at":"2026-07-05T10:19:12.945106+00:00"},{"alias_kind":"pith_short_12","alias_value":"JYTTSXB5SVM2","created_at":"2026-07-05T10:19:12.945106+00:00"},{"alias_kind":"pith_short_16","alias_value":"JYTTSXB5SVM2O5EZ","created_at":"2026-07-05T10:19:12.945106+00:00"},{"alias_kind":"pith_short_8","alias_value":"JYTTSXB5","created_at":"2026-07-05T10:19:12.945106+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.00913","citing_title":"Two AI Metrics Diverged: Will it Make All the Difference?","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03217","citing_title":"An Asymptotic Theory of Chain-of-Thought in In-Context Learning","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28661","citing_title":"When More Sampling Hurts: The Modal Ceiling and Correlation Ceiling of Test-Time Scaling","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07616","citing_title":"Item Response Scaling Laws: A Measurement Theory Approach for Efficient and Generalizable Neural Scaling Estimation","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2510.04265","citing_title":"Don't Pass@k: A Bayesian Framework for Large Language Model Evaluation","ref_index":65,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12412","citing_title":"Stories in Space: In-Context Learning Trajectories in Conceptual Belief Space","ref_index":97,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17614","citing_title":"Characterizing Model-Native Skills","ref_index":93,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17290","citing_title":"Probabilistic Programs of Thought","ref_index":53,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JYTTSXB5SVM2O5EZAR35QLLCN2","json":"https://pith.science/pith/JYTTSXB5SVM2O5EZAR35QLLCN2.json","graph_json":"https://pith.science/api/pith-number/JYTTSXB5SVM2O5EZAR35QLLCN2/graph.json","events_json":"https://pith.science/api/pith-number/JYTTSXB5SVM2O5EZAR35QLLCN2/events.json","paper":"https://pith.science/paper/JYTTSXB5"},"agent_actions":{"view_html":"https://pith.science/pith/JYTTSXB5SVM2O5EZAR35QLLCN2","download_json":"https://pith.science/pith/JYTTSXB5SVM2O5EZAR35QLLCN2.json","view_paper":"https://pith.science/paper/JYTTSXB5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.17578&json=true","fetch_graph":"https://pith.science/api/pith-number/JYTTSXB5SVM2O5EZAR35QLLCN2/graph.json","fetch_events":"https://pith.science/api/pith-number/JYTTSXB5SVM2O5EZAR35QLLCN2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JYTTSXB5SVM2O5EZAR35QLLCN2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JYTTSXB5SVM2O5EZAR35QLLCN2/action/storage_attestation","attest_author":"https://pith.science/pith/JYTTSXB5SVM2O5EZAR35QLLCN2/action/author_attestation","sign_citation":"https://pith.science/pith/JYTTSXB5SVM2O5EZAR35QLLCN2/action/citation_signature","submit_replication":"https://pith.science/pith/JYTTSXB5SVM2O5EZAR35QLLCN2/action/replication_record"}},"created_at":"2026-07-05T10:19:12.945106+00:00","updated_at":"2026-07-05T10:19:12.945106+00:00"}