{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2ALQ5Z2IHCTJ4CXOGDYL3NNTJ2","short_pith_number":"pith:2ALQ5Z2I","schema_version":"1.0","canonical_sha256":"d0170ee74838a69e0aee30f0bdb5b34eb3c7cbebe8cb19e2775f66de34fe47fc","source":{"kind":"arxiv","id":"2502.14382","version":1},"attestation_state":"computed","paper":{"title":"S*: Test Time Scaling for Code Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chengkun Cao, Dacheng Li, Ion Stoica, Jiarong Xing, Joseph E. Gonzalez, Kurt Keutzer, Shangyin Tan, Shiyi Cao, Xiuyu Li","submitted_at":"2025-02-20T09:18:53Z","abstract_excerpt":"Increasing test-time compute for LLMs shows promise across domains but remains underexplored in code generation, despite extensive study in math. In this paper, we propose S*, the first hybrid test-time scaling framework that substantially improves the coverage and selection accuracy of generated code. S* extends the existing parallel scaling paradigm with sequential scaling to push performance boundaries. It further leverages a novel selection mechanism that adaptively generates distinguishing inputs for pairwise comparison, combined with execution-grounded information to robustly identify co"},"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":"2502.14382","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-20T09:18:53Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"7f6a404271d08599c85f5a69adba5d0b7b761c55d96f0c830845cacbd85986ae","abstract_canon_sha256":"f52857622fd163943381389fe908164be603e666a7d2f5a2cdf26c986bff97c0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:17:25.534135Z","signature_b64":"CHOo+NpPjCQAe/JbREEXRaIC6yM0y4LYkkbmJlt4eQ9Thb+HhCxSDBKsHdQRb1hE0yIHtpyoM7//gLxDK2wiDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d0170ee74838a69e0aee30f0bdb5b34eb3c7cbebe8cb19e2775f66de34fe47fc","last_reissued_at":"2026-07-05T10:17:25.533613Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:17:25.533613Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"S*: Test Time Scaling for Code Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chengkun Cao, Dacheng Li, Ion Stoica, Jiarong Xing, Joseph E. Gonzalez, Kurt Keutzer, Shangyin Tan, Shiyi Cao, Xiuyu Li","submitted_at":"2025-02-20T09:18:53Z","abstract_excerpt":"Increasing test-time compute for LLMs shows promise across domains but remains underexplored in code generation, despite extensive study in math. In this paper, we propose S*, the first hybrid test-time scaling framework that substantially improves the coverage and selection accuracy of generated code. S* extends the existing parallel scaling paradigm with sequential scaling to push performance boundaries. It further leverages a novel selection mechanism that adaptively generates distinguishing inputs for pairwise comparison, combined with execution-grounded information to robustly identify co"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.14382","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/2502.14382/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":"2502.14382","created_at":"2026-07-05T10:17:25.533680+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.14382v1","created_at":"2026-07-05T10:17:25.533680+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.14382","created_at":"2026-07-05T10:17:25.533680+00:00"},{"alias_kind":"pith_short_12","alias_value":"2ALQ5Z2IHCTJ","created_at":"2026-07-05T10:17:25.533680+00:00"},{"alias_kind":"pith_short_16","alias_value":"2ALQ5Z2IHCTJ4CXO","created_at":"2026-07-05T10:17:25.533680+00:00"},{"alias_kind":"pith_short_8","alias_value":"2ALQ5Z2I","created_at":"2026-07-05T10:17:25.533680+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":18,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.27268","citing_title":"E-TTS: A New Embodied Test-Time Scaling Framework for Robotic Manipulation","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17682","citing_title":"From Trainee to Trainer: LLM-Designed Training Environment for RL with Multi-Agent Reasoning","ref_index":65,"is_internal_anchor":false},{"citing_arxiv_id":"2606.11052","citing_title":"Attention Amnesia in Hybrid LLMs: When CoT Fine-Tuning Breaks Long-Range Recall, and How to Fix It","ref_index":65,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03102","citing_title":"Small RL Controller, Large Language Model: RL-Guided Adaptive Sampling for Test-Time Scaling","ref_index":96,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22478","citing_title":"DeliCIR: Memory-Guided Test-Time Deliberation via Multi-Agent Collaboration for Composed Image Retrieval","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22478","citing_title":"DeliCIR: Memory-Guided Test-Time Deliberation via Multi-Agent Collaboration for Composed Image Retrieval","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2511.16858","citing_title":"Investigating Test Overfitting on SWE-bench","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2512.14917","citing_title":"Evaluating Code Reasoning Abilities of Large Language Models Under Real-World Settings","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2603.27098","citing_title":"Ensemble-Based Uncertainty Estimation for Code Correctness Estimation","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11299","citing_title":"Primal Generation, Dual Judgment: Self-Training from Test-Time Scaling","ref_index":63,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11299","citing_title":"Primal Generation, Dual Judgment: Self-Training from Test-Time Scaling","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01194","citing_title":"VLA-ATTC: Adaptive Test-Time Compute for VLA Models with Relative Action Critic Model","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10449","citing_title":"AdverMCTS: Combating Pseudo-Correctness in Code Generation via Adversarial Monte Carlo Tree Search","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2604.05868","citing_title":"Understanding Performance Gap Between Parallel and Sequential Sampling in Large Reasoning Models","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2604.05560","citing_title":"An Iterative Test-and-Repair Framework for Competitive Code Generation","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16995","citing_title":"SPS: Steering Probability Squeezing for Better Exploration in Reinforcement Learning for Large Language Models","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17288","citing_title":"Clover: A Neural-Symbolic Agentic Harness with Stochastic Tree-of-Thoughts for Verified RTL Repair","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21598","citing_title":"You Don't Need Public Tests to Generate Correct Code","ref_index":12,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2ALQ5Z2IHCTJ4CXOGDYL3NNTJ2","json":"https://pith.science/pith/2ALQ5Z2IHCTJ4CXOGDYL3NNTJ2.json","graph_json":"https://pith.science/api/pith-number/2ALQ5Z2IHCTJ4CXOGDYL3NNTJ2/graph.json","events_json":"https://pith.science/api/pith-number/2ALQ5Z2IHCTJ4CXOGDYL3NNTJ2/events.json","paper":"https://pith.science/paper/2ALQ5Z2I"},"agent_actions":{"view_html":"https://pith.science/pith/2ALQ5Z2IHCTJ4CXOGDYL3NNTJ2","download_json":"https://pith.science/pith/2ALQ5Z2IHCTJ4CXOGDYL3NNTJ2.json","view_paper":"https://pith.science/paper/2ALQ5Z2I","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.14382&json=true","fetch_graph":"https://pith.science/api/pith-number/2ALQ5Z2IHCTJ4CXOGDYL3NNTJ2/graph.json","fetch_events":"https://pith.science/api/pith-number/2ALQ5Z2IHCTJ4CXOGDYL3NNTJ2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2ALQ5Z2IHCTJ4CXOGDYL3NNTJ2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2ALQ5Z2IHCTJ4CXOGDYL3NNTJ2/action/storage_attestation","attest_author":"https://pith.science/pith/2ALQ5Z2IHCTJ4CXOGDYL3NNTJ2/action/author_attestation","sign_citation":"https://pith.science/pith/2ALQ5Z2IHCTJ4CXOGDYL3NNTJ2/action/citation_signature","submit_replication":"https://pith.science/pith/2ALQ5Z2IHCTJ4CXOGDYL3NNTJ2/action/replication_record"}},"created_at":"2026-07-05T10:17:25.533680+00:00","updated_at":"2026-07-05T10:17:25.533680+00:00"}