{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.10366", "kind": "paper", "url": "https://arxiv.org/abs/2608.10366"}, {"id": "gh:vis-nlp/dsagentbench", "kind": "repo", "url": "https://github.com/vis-nlp/DSAgentBench"}], "authors": ["Mizanur Rahman", "Mohammed Saidul Islam", "Ridwan Mahbub", "Md Tahmid Rahman Laskar", "Shafiq Joty", "Enamul Hoque Prince"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.10366", "source_url": "https://arxiv.org/abs/2608.10366"}, "categories": ["General AI", "Language & Knowledge", "cs.AI"], "description": {"en": "DSAgentBench evaluates language agents on end-to-end data-science workflows in real computer environments. It comprises 275 tasks spanning data wrangling, exploration, modeling, visualization, and validation, with deterministic evaluators verifying analytical correctness, visual outputs, and model performance."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.10366", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.10366", "title": "DSAgentBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.10366", "modality": null, "name": "DSAgentBench", "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_dsagentbench_96674f61", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "a584f4f70589bf72cb22ce824cf3fe9e1721dcb9a71586a2df345b7975812eb6", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2608.10366"}, "publisher": null, "released": "2026-08-11", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.10366", "source_url": "https://arxiv.org/abs/2608.10366"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-10366", "source": "claire_radar", "source_benchmark_id": "2608.10366", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Language & Knowledge", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/vis-nlp/DSAgentBench", "githubScope": "benchmark_repo", "githubStars": 4, "hfDailySubmittedAt": "2026-08-12T00:00:00.000Z", "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 11, "hfPaperUrl": "https://huggingface.co/papers/2608.10366", "observedAt": "2026-10-04T15:31:48.030916Z", "signalStatus": {"githubStars": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}}, "status": "fresh"}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "13a48bd9c8ed99ab691727a6ba073787497b8648c31139f631ca072d3c8ca472", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "DSAgentBench evaluates language agents on end-to-end data-science workflows in real computer environments. It comprises 275 tasks spanning data wrangling, exploration, modeling, visualization, and validation, with deterministic evaluators verifying analytical correctness, visual outputs, and model performance.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "We introduce DSAgentBench, the first benchmark to evaluate whether agents can automate full data-science workflows inside real computer environments."}, "familyId": "bmf_3b3386cca544", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [{"evidence": [{"basis": "existing-code-link", "linkedUrl": "https://github.com/vis-nlp/DSAgentBench", "url": "https://github.com/vis-nlp/DSAgentBench"}, {"basis": "existing-metric-association", "linkedUrl": "https://github.com/vis-nlp/DSAgentBench", "url": "https://huggingface.co/papers/2608.10366"}, {"basis": "huggingface-paper-association", "linkedUrl": "https://github.com/vis-nlp/DSAgentBench", "url": "https://huggingface.co/api/papers/2608.10366"}, {"basis": "linked-from-report", "linkedUrl": "https://github.com/vis-nlp/DSAgentBench", "url": "https://arxiv.org/abs/2608.10366"}, {"basis": "linked-from-report", "linkedUrl": "https://github.com/vis-nlp/DSAgentBench.", "url": "https://arxiv.org/abs/2608.10366"}], "scope": "benchmark_repo", "url": "https://github.com/vis-nlp/DSAgentBench", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_dsagentbench_96674f61", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/vis-nlp/DSAgentBench", "data": null, "hfPaper": "https://huggingface.co/papers/2608.10366", "paper": "https://arxiv.org/abs/2608.10366", "pdf": "https://arxiv.org/pdf/2608.10366", "project": null, "report": "https://arxiv.org/abs/2608.10366"}, "metrics": [], "motivation": "Real-world data science involves long-horizon workflows that span data wrangling, exploration, modeling, visualization, and validation, and require coordinated use of tools such as notebooks, IDEs, terminals, browsers, and databases within real operating environments.", "name": "DSAgentBench", "oneLine": "DSAgentBench evaluates language agents on end-to-end data-science workflows in real computer environments. It comprises 275 tasks spanning data wrangling, exploration, modeling, visualization, and validation, with deterministic evaluators verifying analytical correctness, visual outputs, and model performance.", "paperTitle": "DSAgentBench: Can Agents Automate End-to-End Data-Science Workflows in Real Computer Environments?", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.10366", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "vis-nlp", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://github.com/vis-nlp/DSAgentBench"}], "ranking": {"90d": {"confidence": "Medium", "coverage": 0.7, "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": 0.21483102985913266, "value": 4}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.4390696100905538, "value": 11}}, "confidence": "Medium", "coverage": 0.7, "rank": 208, "score": 33}, "method": "attention-ranking-v13", "rank": 208, "score": 33}}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-08-11", "source": {"authors": ["Mizanur Rahman", "Mohammed Saidul Islam", "Ridwan Mahbub", "Md Tahmid Rahman Laskar", "Shafiq Joty", "Enamul Hoque Prince"], "categories": ["cs.AI", "cs.CL"], "id": "2608.10366", "title": "DSAgentBench: Can Agents Automate End-to-End Data-Science Workflows in Real Computer Environments?", "type": "arxiv", "url": "https://arxiv.org/abs/2608.10366"}, "sourceUpdatedAt": "2026-08-11T01:45:56Z", "topics": ["cs.AI"], "venueAttempts": [], "whyItMatters": "Existing benchmarks lack real-computer interaction and fail to capture the multi-stage, multi-tool nature of data-science practice. DSAgentBench provides a realistic environment for assessing whether agents can automate complete workflows, highlighting a significant capability gap and offering a foundation for developing grounded, verifiable autonomous agents."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
