{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2607.01597", "kind": "paper", "url": "https://arxiv.org/abs/2607.01597"}], "authors": ["Boyang Yang", "Xiangliang Hu", "Luyao Ren", "Yanjun Chen", "Bach Le", "Tegawendé F. Bissyandé", "Haoye Tian"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2607.01597", "source_url": "https://arxiv.org/abs/2607.01597"}, "categories": ["Code & Software", "General AI", "cs.SE"], "description": {"en": "PatchFuseBench is a fixed-pool benchmark for evaluating repair candidate fusion, built from existing SWE-bench Verified, SWE-bench Multilingual, and Defects4J candidate patches. The benchmark pools candidate patches for 500 bugs on SWE-bench Verified, 300 on Multilingual, and 371 on Defects4J, and evaluates methods that fuse or select a final patch."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2607.01597", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2607.01597", "title": "PatchFuseBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Yanshan University", "Peking University", "Google (United States)", "The University of Melbourne", "University of Luxembourg", "Aalto University"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2607.01597", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2607.01597", "source_url": "https://openalex.org/W7167214889"}, "key": "claire-radar:2607.01597", "modality": null, "name": "PatchFuseBench", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "Medium", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "unknown", "origin_record_id": "bm_patchfusebench_d8b0e871", "origin_source": "arxiv", "recognition_confidence": "0.85", "record_sha256": "9b46cb98cd54cce3a3c5cfab79869ebdfdbd2056c57a4f553b43d58910587da8", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2607.01597"}, "publisher": null, "released": "2026-07-02", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2607.01597", "source_url": "https://arxiv.org/abs/2607.01597"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2607-01597", "source": "claire_radar", "source_benchmark_id": "2607.01597", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Code & Software", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2607.01597", "observedAt": "2026-10-04T03:02:42.754940Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "state": "unavailable"}}, "status": "unavailable"}, "capabilities": [], "confidence": "Medium", "construction": "Transform Existing", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "a28f0649ea596e6fecd5fc741ddafdd391dd31066feb15042512db4c67884169", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "PatchFuseBench is a fixed-pool benchmark for evaluating repair candidate fusion, built from existing SWE-bench Verified, SWE-bench Multilingual, and Defects4J candidate patches. The benchmark pools candidate patches for 500 bugs on SWE-bench Verified, 300 on Multilingual, and 371 on Defects4J, and evaluates methods that fuse or select a final patch.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact named benchmark artifact released in abstract", "benchmark release stated in one sentence"], "snippet": "To evaluate this setting, we build PatchFuseBench, a fixed-pool benchmark covering SWE-bench Verified, SWE-bench Multilingual, and Defects4J candidate patches."}, "familyId": "bmf_060068620f54", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_patchfusebench_d8b0e871", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2607.01597", "paper": "https://arxiv.org/abs/2607.01597", "pdf": "https://arxiv.org/pdf/2607.01597", "project": null, "report": "https://arxiv.org/abs/2607.01597"}, "metrics": [], "motivation": "Modern LLM coding agents are commonly evaluated using pass@k, but developers typically apply a single final patch in real-world settings.", "name": "PatchFuseBench", "oneLine": "PatchFuseBench is a fixed-pool benchmark for evaluating repair candidate fusion, built from existing SWE-bench Verified, SWE-bench Multilingual, and Defects4J candidate patches. The benchmark pools candidate patches for 500 bugs on SWE-bench Verified, 300 on Multilingual, and 371 on Defects4J, and evaluates methods that fuse or select a final patch.", "paperTitle": "A Single Patch Is Not Enough: Deterministic Fusion of Repair Candidates", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2607.01597", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 0.85, "relation": "introduces", "releasedAt": "2026-07-02", "source": {"authors": ["Boyang Yang", "Xiangliang Hu", "Luyao Ren", "Yanjun Chen", "Bach Le", "Tegawend\\'e F. Bissyand\\'e", "Haoye Tian"], "categories": ["cs.SE"], "id": "2607.01597", "title": "A Single Patch Is Not Enough: Deterministic Fusion of Repair Candidates", "type": "arxiv", "url": "https://arxiv.org/abs/2607.01597"}, "sourceUpdatedAt": "2026-07-02T01:53:17Z", "topics": ["cs.SE"], "venueAttempts": [], "whyItMatters": "It addresses the pass@k-to-pass@1 gap in code repair, where candidate pools may contain correct patches but selection remains challenging. The benchmark provides a controlled setting to compare post-generation patch selection and fusion methods."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
