{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.09802", "kind": "paper", "url": "https://arxiv.org/abs/2608.09802"}, {"id": "hf:swe-bench-promax/swe-bench-promax", "kind": "dataset", "url": "https://huggingface.co/datasets/swe-bench-promax/SWE-Bench-ProMax"}], "authors": ["Yuling Shi", "Jinghan Xu", "Kelin Fu", "Wenhao Zeng", "Shilin He", "Lei Zhang", "Yue Liu", "Zelin Zhao", "Terry Yue Zhuo", "Jialun Cao", "Siyu Ye", "Tianyu Liu", "Kai Cai", "Shing-Chi Cheung", "Xiaodong Gu"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.09802", "source_url": "https://arxiv.org/abs/2608.09802"}, "categories": ["Code & Software", "Code generation", "Software & AI Compute", "cs.CL"], "description": {"en": "A multilingual code refactoring benchmark with 170 instances drawn from real commits across seven programming languages. Evaluates AI agents on large-scale refactoring tasks averaging 11.4 modified files and 261.6 lines of code, using manually curated issue descriptions and test suites."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.09802", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.09802", "title": "SWE-Bench ProMax"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Shanghai Jiao Tong University", "Peking University", "University of Chinese Academy of Sciences", "National University of Singapore", "Monash University", "Hong Kong University of Science and Technology"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2608.09802", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2608.09802", "source_url": "https://openalex.org/W7202186081"}, "key": "claire-radar:2608.09802", "modality": null, "name": "SWE-Bench ProMax", "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_swe-bench_6afbfb26", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "fa0ea4ff67611280f7e0e083fe0b5e88f1a60a0b51e0cbabcd934175de3c4011", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2608.09802"}, "publisher": null, "released": "2026-08-10", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.09802", "source_url": "https://arxiv.org/abs/2608.09802"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-09802", "source": "claire_radar", "source_benchmark_id": "2608.09802", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["Software & AI Compute"], "area": "Code & Software", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": "2026-08-11T00:00:00.000Z", "hfDataset": "swe-bench-promax/SWE-Bench-ProMax", "hfDatasetDownloads": 807, "hfDatasetLikes": 1, "hfPaperUpvotes": 84, "hfPaperUrl": "https://huggingface.co/papers/2608.09802", "observedAt": "2026-10-04T15:31:48.030916Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}, "hfDatasetLikes": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}, "hfPaperUpvotes": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}}, "status": "fresh"}, "capabilities": ["Code generation"], "confidence": "High", "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": "fa4c4a74da32976391c428ed62af680b43ba9fd7cdf6d17b40d422e01bcd079f", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "A multilingual code refactoring benchmark with 170 instances drawn from real commits across seven programming languages. Evaluates AI agents on large-scale refactoring tasks averaging 11.4 modified files and 261.6 lines of code, using manually curated issue descriptions and test suites.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact named benchmark artifact released in abstract", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "We introduce SWE-Bench ProMax, an expert-curated, multilingual code refactoring benchmark of 170 instances drawn from real commits across seven programming languages (Python, Java, TypeScript, Go, C, C++, and Rust)."}, "familyId": "bmf_7e98773a0ce5", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_swe-bench_6afbfb26", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": ["Software & Cloud"], "links": {"code": null, "data": "https://huggingface.co/datasets/swe-bench-promax/SWE-Bench-ProMax", "hfPaper": "https://huggingface.co/papers/2608.09802", "paper": "https://arxiv.org/abs/2608.09802", "pdf": "https://arxiv.org/pdf/2608.09802", "project": null, "report": "https://arxiv.org/abs/2608.09802"}, "metrics": [], "motivation": "As AI coding agents take on increasingly complex, long-horizon software engineering tasks, existing benchmarks are rapidly saturating and their evaluation quality has come under serious scrutiny: a recent audit found that nearly 60% of unsolved SWE-bench Verified instances contain flawed tests -- either overly narrow tests that reject correct solutions or overly broad tests that check unstated requirements -- and th…", "name": "SWE-Bench ProMax", "oneLine": "A multilingual code refactoring benchmark with 170 instances drawn from real commits across seven programming languages. Evaluates AI agents on large-scale refactoring tasks averaging 11.4 modified files and 261.6 lines of code, using manually curated issue descriptions and test suites.", "paperTitle": "SWE-Bench ProMax: Benchmarking Agents on Large-Scale Multilingual Code Refactoring", "primaryDomain": "Software & AI Compute", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.09802", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "SWE-Bench-ProMax Team", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://huggingface.co/datasets/swe-bench-promax/SWE-Bench-ProMax"}], "ranking": {"90d": {"confidence": "Medium", "coverage": 0.45, "datasetDownloadRank": 22, "datasetRankPopulation": 85, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": null, "value": null}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": 0.6894054845250103, "value": 807}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.7849925288428563, "value": 84}}, "confidence": "Medium", "coverage": 0.45, "rank": 19, "score": 60}, "method": "attention-ranking-v13", "rank": 19, "score": 60}}, "readiness": "Inspectable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-08-10", "source": {"authors": ["Yuling Shi", "Jinghan Xu", "Kelin Fu", "Wenhao Zeng", "Shilin He", "Lei Zhang", "Yue Liu", "Zelin Zhao", "Terry Yue Zhuo", "Jialun Cao", "Siyu Ye", "Tianyu Liu", "Kai Cai", "Shing-Chi Cheung", "Xiaodong Gu"], "categories": ["cs.CL", "cs.SE"], "id": "2608.09802", "title": "SWE-Bench ProMax: Benchmarking Agents on Large-Scale Multilingual Code Refactoring", "type": "arxiv", "url": "https://arxiv.org/abs/2608.09802"}, "sourceUpdatedAt": "2026-08-10T16:23:19Z", "topics": ["cs.CL"], "venueAttempts": [], "whyItMatters": "Existing software engineering benchmarks face saturation and quality issues, with flawed tests and training data leakage. This benchmark provides a more challenging and realistic refactoring task set with rigorous curation, offering a robust measure of agent capability for long-horizon coding tasks."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
