{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2605.29054", "kind": "paper", "url": "https://arxiv.org/abs/2605.29054"}], "authors": ["Tao Feng", "Jiefeng Chen", "Linxin Song", "Yufan Zhuang", "Yue Huang", "Bhavana Dalvi Mishra", "Chi Wang", "Xiangru Tang", "Jieyu Zhao", "Ge Liu", "Tomas Pfister", "Jinsung Yoon"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2605.29054", "source_url": "https://arxiv.org/abs/2605.29054"}, "categories": ["Code & Software", "General AI", "cs.SE"], "description": {"en": "T2J-Bench benchmarks codebase conversion by transferring PyTorch code to JAX under a fixed equivalence contract with three ordered verification stages."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2605.29054", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2605.29054", "title": "T2J-Bench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Google (United States)", "University of Southern California", "University of Notre Dame"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2605.29054", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2605.29054", "source_url": "https://openalex.org/W7162767610"}, "key": "claire-radar:2605.29054", "modality": null, "name": "T2J-Bench", "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_t2j-bench_85fea4b8", "origin_source": "arxiv", "recognition_confidence": "0.85", "record_sha256": "b043ca38078ba446bd0e81f5bac8807ec2c029abea85c5ceea0647b08d132434", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2605.29054"}, "publisher": null, "released": "2026-05-27", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2605.29054", "source_url": "https://arxiv.org/abs/2605.29054"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2605-29054", "source": "claire_radar", "source_benchmark_id": "2605.29054", "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": 0, "hfPaperUrl": "https://huggingface.co/papers/2605.29054", "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", "lastSuccessfulDate": "2026-09-30", "state": "stale"}}, "status": "stale"}, "capabilities": [], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "ab69166eb290d1e70f471e6c40ad2ace2ebbfc098fc1b206de14ef93048de688", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "T2J-Bench benchmarks codebase conversion by transferring PyTorch code to JAX under a fixed equivalence contract with three ordered verification stages.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact named benchmark artifact released in abstract", "benchmark release stated in one sentence"], "snippet": "We introduce T2J-Bench, a benchmark for codebase conversion that reformulates conversion as transfer under a fixed equivalence contract."}, "familyId": "bmf_92b4500c07b1", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_t2j-bench_85fea4b8", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2605.29054", "paper": "https://arxiv.org/abs/2605.29054", "pdf": "https://arxiv.org/pdf/2605.29054", "project": null, "report": "https://arxiv.org/abs/2605.29054"}, "metrics": [], "motivation": "Coding agents increasingly act as codebase-scale collaborators that can assist with codebase conversion, but this progress has exposed a critical weakness: agents often over-trust their own local validation routines and declare success on artifacts that satisfy surface checks while violating the semantic contracts users actually care about.", "name": "T2J-Bench", "oneLine": "T2J-Bench benchmarks codebase conversion by transferring PyTorch code to JAX under a fixed equivalence contract with three ordered verification stages.", "paperTitle": "Converted, Not Equivalent: Benchmarking Codebase Conversion via Observational Equivalence", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2605.29054", "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": "2026-05-27", "source": {"authors": ["Linxin Song", "Jiefeng Chen", "Yue Huang", "Bhavana Dalvi Mishra", "Chi Wang", "Jieyu Zhao", "Jinsung Yoon", "Tomas Pfister"], "categories": ["cs.SE", "cs.CL"], "id": "2605.29054", "title": "Converted, Not Equivalent: Benchmarking Codebase Conversion via Observational Equivalence", "type": "arxiv", "url": "https://arxiv.org/abs/2605.29054"}, "sourceUpdatedAt": "2026-06-03T18:44:55Z", "topics": ["cs.SE"], "venueAttempts": [], "whyItMatters": "Reveals that agents overestimate success on codebase conversion, but the benchmark's data and verification harness are not publicly released."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
