{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:C2OBP3ZJN6ICXGTBYRYPSVZMYI","short_pith_number":"pith:C2OBP3ZJ","schema_version":"1.0","canonical_sha256":"169c17ef296f902b9a61c470f9572cc226c33ef424f7273307e978f6be94ae85","source":{"kind":"arxiv","id":"2501.14723","version":2},"attestation_state":"computed","paper":{"title":"CodeMonkeys: Scaling Test-Time Compute for Software Engineering","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Azalia Mirhoseini, Bradley Brown, Christopher R\\'e, Jordan Juravsky, Ronald Clark, Ryan Ehrlich","submitted_at":"2025-01-24T18:58:40Z","abstract_excerpt":"Scaling test-time compute is a promising axis for improving LLM capabilities. However, test-time compute can be scaled in a variety of ways, and effectively combining different approaches remains an active area of research. Here, we explore this problem in the context of solving real-world GitHub issues from the SWE-bench dataset. Our system, named CodeMonkeys, allows models to iteratively edit a codebase by jointly generating and running a testing script alongside their draft edit. We sample many of these multi-turn trajectories for every issue to generate a collection of candidate edits. Thi"},"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":"2501.14723","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-24T18:58:40Z","cross_cats_sorted":[],"title_canon_sha256":"c8f87ad12e60940bdeeb77c5a86e75a083d768ce42167a8474997188123920d3","abstract_canon_sha256":"b3b798154f9310add1a74dc7e47b7ccc50e34e99142c00cbca0294a2b3858134"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:08:54.510514Z","signature_b64":"KRqmnYIc8A5LBb+KNgraneEOWlVk/Oad/5AAsyAX6J2kLcr4Ga0+2HZN7m7T3s9zUUWvcBqLRLCEZgrnD++QAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"169c17ef296f902b9a61c470f9572cc226c33ef424f7273307e978f6be94ae85","last_reissued_at":"2026-07-05T10:08:54.510028Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:08:54.510028Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CodeMonkeys: Scaling Test-Time Compute for Software Engineering","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Azalia Mirhoseini, Bradley Brown, Christopher R\\'e, Jordan Juravsky, Ronald Clark, Ryan Ehrlich","submitted_at":"2025-01-24T18:58:40Z","abstract_excerpt":"Scaling test-time compute is a promising axis for improving LLM capabilities. However, test-time compute can be scaled in a variety of ways, and effectively combining different approaches remains an active area of research. Here, we explore this problem in the context of solving real-world GitHub issues from the SWE-bench dataset. Our system, named CodeMonkeys, allows models to iteratively edit a codebase by jointly generating and running a testing script alongside their draft edit. We sample many of these multi-turn trajectories for every issue to generate a collection of candidate edits. Thi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.14723","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/2501.14723/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":"2501.14723","created_at":"2026-07-05T10:08:54.510083+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.14723v2","created_at":"2026-07-05T10:08:54.510083+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.14723","created_at":"2026-07-05T10:08:54.510083+00:00"},{"alias_kind":"pith_short_12","alias_value":"C2OBP3ZJN6IC","created_at":"2026-07-05T10:08:54.510083+00:00"},{"alias_kind":"pith_short_16","alias_value":"C2OBP3ZJN6ICXGTB","created_at":"2026-07-05T10:08:54.510083+00:00"},{"alias_kind":"pith_short_8","alias_value":"C2OBP3ZJ","created_at":"2026-07-05T10:08:54.510083+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":13,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.26173","citing_title":"AlgoEvolve: LLM-driven Meta-evolution of Algorithmic Trading Programs","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2606.21228","citing_title":"Sakana Fugu Technical Report","ref_index":179,"is_internal_anchor":false},{"citing_arxiv_id":"2607.02032","citing_title":"PACE: A Proxy for Agentic Capability Evaluation","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2607.01597","citing_title":"A Single Patch Is Not Enough: Deterministic Fusion of Repair Candidates","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15669","citing_title":"Rule2DRC: Benchmarking LLM Agents for DRC Script Synthesis with Execution-Guided Test Generation","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2510.18270","citing_title":"Can Old Tests Do New Tricks for Resolving SWE Issues?","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2511.16858","citing_title":"Investigating Test Overfitting on SWE-bench","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2512.06443","citing_title":"Vec-LUT: Vector Table Lookup for Parallel Ultra-Low-Bit LLM Inference on Edge Devices","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2509.19349","citing_title":"ShinkaEvolve: Towards Open-Ended And Sample-Efficient Program Evolution","ref_index":145,"is_internal_anchor":false},{"citing_arxiv_id":"2602.07900","citing_title":"Rethinking the Value of Agent-Generated Tests for LLM-Based Software Engineering Agents","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04320","citing_title":"Reproduction Test Generation for Java SWE Issues","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04320","citing_title":"Reproduction Test Generation for Java SWE Issues","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17433","citing_title":"Self-Consistency from Only Two Samples: CoT-PoT Ensembling for Efficient LLM Reasoning","ref_index":42,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/C2OBP3ZJN6ICXGTBYRYPSVZMYI","json":"https://pith.science/pith/C2OBP3ZJN6ICXGTBYRYPSVZMYI.json","graph_json":"https://pith.science/api/pith-number/C2OBP3ZJN6ICXGTBYRYPSVZMYI/graph.json","events_json":"https://pith.science/api/pith-number/C2OBP3ZJN6ICXGTBYRYPSVZMYI/events.json","paper":"https://pith.science/paper/C2OBP3ZJ"},"agent_actions":{"view_html":"https://pith.science/pith/C2OBP3ZJN6ICXGTBYRYPSVZMYI","download_json":"https://pith.science/pith/C2OBP3ZJN6ICXGTBYRYPSVZMYI.json","view_paper":"https://pith.science/paper/C2OBP3ZJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.14723&json=true","fetch_graph":"https://pith.science/api/pith-number/C2OBP3ZJN6ICXGTBYRYPSVZMYI/graph.json","fetch_events":"https://pith.science/api/pith-number/C2OBP3ZJN6ICXGTBYRYPSVZMYI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/C2OBP3ZJN6ICXGTBYRYPSVZMYI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/C2OBP3ZJN6ICXGTBYRYPSVZMYI/action/storage_attestation","attest_author":"https://pith.science/pith/C2OBP3ZJN6ICXGTBYRYPSVZMYI/action/author_attestation","sign_citation":"https://pith.science/pith/C2OBP3ZJN6ICXGTBYRYPSVZMYI/action/citation_signature","submit_replication":"https://pith.science/pith/C2OBP3ZJN6ICXGTBYRYPSVZMYI/action/replication_record"}},"created_at":"2026-07-05T10:08:54.510083+00:00","updated_at":"2026-07-05T10:08:54.510083+00:00"}