{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2607.03523", "kind": "paper", "url": "https://arxiv.org/abs/2607.03523"}], "authors": ["Caroline Choi", "Zeyneb Kaya", "Shirley Wu", "Tengyu Ma", "Tatsunori Hashimoto", "Ludwig Schmidt"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2607.03523", "source_url": "https://arxiv.org/abs/2607.03523"}, "categories": ["Code & Software", "General AI", "cs.SE"], "description": {"en": "The work introduces BugSourceBench, a code repair benchmark with bugs from human-written, LM-generated, and human-edited LM-generated code. The benchmark evaluates the fix rate of language models on these bug sources. No scoring contract, dataset, or public access path is provided."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2607.03523", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2607.03523", "title": "BugSourceBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2607.03523", "modality": null, "name": "BugSourceBench", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "High", "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_bugsourcebench_7e234238", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "a0e62226ab360bca5e828cc02e2b919b24629e5a6855e1b1fc4bcc1be08be39f", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2607.03523"}, "publisher": null, "released": "2026-07-03", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2607.03523", "source_url": "https://arxiv.org/abs/2607.03523"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2607-03523", "source": "claire_radar", "source_benchmark_id": "2607.03523", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Machine Generated", "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/2607.03523", "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": "High", "construction": "Original Synthetic", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "b9c19a3411cf118e7b6ffe7f130362be00567913ec033ad6be4d7c635e51121f", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "The work introduces BugSourceBench, a code repair benchmark with bugs from human-written, LM-generated, and human-edited LM-generated code. The benchmark evaluates the fix rate of language models on these bug sources. No scoring contract, dataset, or public access path is provided.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact named benchmark artifact released in abstract", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "To test whether this curriculum generalizes, we introduce BugSourceBench, a repair benchmark spanning realistic bug sources: bugs in human-written code, LM-generated code, and human-edited LM-generated code."}, "familyId": "bmf_7f56fcdc5c5d", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_bugsourcebench_7e234238", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2607.03523", "paper": "https://arxiv.org/abs/2607.03523", "pdf": "https://arxiv.org/pdf/2607.03523", "project": null, "report": "https://arxiv.org/abs/2607.03523"}, "metrics": [], "motivation": "Code repair is an important capability for language models (LMs): given a buggy program and unit tests, an LM must produce a fixed program that passes the tests.", "name": "BugSourceBench", "oneLine": "The work introduces BugSourceBench, a code repair benchmark with bugs from human-written, LM-generated, and human-edited LM-generated code. The benchmark evaluates the fix rate of language models on these bug sources. No scoring contract, dataset, or public access path is provided.", "paperTitle": "Anchored Self-Play for Code Repair", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2607.03523", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-07-03", "source": {"authors": ["Caroline Choi", "Zeyneb Kaya", "Shirley Wu", "Tengyu Ma", "Tatsunori Hashimoto", "Ludwig Schmidt"], "categories": ["cs.SE", "cs.CL"], "id": "2607.03523", "title": "Anchored Self-Play for Code Repair", "type": "arxiv", "url": "https://arxiv.org/abs/2607.03523"}, "sourceUpdatedAt": "2026-07-03T17:51:52Z", "topics": ["cs.SE"], "venueAttempts": [], "whyItMatters": "Code repair benchmarks often focus on synthetic bugs, which may not reflect real-world failures. BugSourceBench aims to cover diverse bug origins, potentially offering a more realistic evaluation for repair models. However, the lack of accessible artifacts and evaluation protocol limits its current utility."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
