{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.16826", "kind": "paper", "url": "https://arxiv.org/abs/2606.16826"}, {"kind": "website", "url": "https://flageval-baai.github.io/AtomBenchPage"}], "authors": ["Zenan Wu", "Bingqing Wei", "Lu Liu", "Zheqi He", "Xi Wang", "Jiakang Liu", "Zehui Li", "Guocai Yao", "Jing-Shu Zheng", "Xi Yang", "Yongtao Wang"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.16826", "source_url": "https://arxiv.org/abs/2606.16826"}, "categories": ["Robot manipulation", "Robotics", "Robotics & Embodied AI"], "description": {"en": "ATOM-Bench evaluates atomic skills and compositional generalization in manipulation policies across 30 atomic tasks and 24 held-out compositional tasks, using paired single-arm and dual-arm robot tracks, with 3,000 human demonstrations and evaluation rollout data released."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.16826", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.16826", "title": "ATOM-Bench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2606.16826", "modality": null, "name": "ATOM-Bench", "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_atom-bench_82c6474b", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "979f8967398a91963335147a5c8bf6722aa6c9154b6f4fbce2800738fab85131", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.16826"}, "publisher": null, "released": "2026-06-15", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.16826", "source_url": "https://arxiv.org/abs/2606.16826"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-16826", "source": "claire_radar", "source_benchmark_id": "2606.16826", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["Robotics & Embodied AI"], "area": "Robotics & Embodied AI", "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/2606.16826", "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": ["Robot manipulation"], "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": "a7d424463565a060cc14b7d35ed2016941916957d47ea8ff6f0f77684b046304", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "ATOM-Bench evaluates atomic skills and compositional generalization in manipulation policies across 30 atomic tasks and 24 held-out compositional tasks, using paired single-arm and dual-arm robot tracks, with 3,000 human demonstrations and evaluation rollout data released.", "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 \\textbf{ATOM-Bench}, a real-world benchmark for evaluating both atomic skills and compositional generalization in manipulation policies."}, "familyId": "bmf_728bb9ed779a", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_atom-bench_82c6474b", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": ["Robotics"], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2606.16826", "paper": "https://arxiv.org/abs/2606.16826", "pdf": "https://arxiv.org/pdf/2606.16826", "project": "https://flageval-baai.github.io/AtomBenchPage", "report": "https://arxiv.org/abs/2606.16826"}, "metrics": [], "motivation": "Generalist manipulation policies are increasingly presented as foundation models for robotic control, but their real-world generalization remains difficult to diagnose.", "name": "ATOM-Bench", "oneLine": "ATOM-Bench evaluates atomic skills and compositional generalization in manipulation policies across 30 atomic tasks and 24 held-out compositional tasks, using paired single-arm and dual-arm robot tracks, with 3,000 human demonstrations and evaluation rollout data released.", "paperTitle": "ATOM-Bench: A Real-World Benchmark for Atomic Skills and Compositional Generalization in Manipulation Policies", "primaryDomain": "Robotics & Embodied AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.16826", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "BAAI", "organizationType": "company-research-lab", "role": "benchmark-publisher", "sourceUrl": "https://flageval-baai.github.io/AtomBenchPage"}], "ranking": {}, "readiness": "Inspectable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-06-15", "source": {"authors": ["Zenan Wu", "Bingqing Wei", "Lu Liu", "Zheqi He", "Xi Wang", "Jiakang Liu", "Zehui Li", "Guocai Yao", "Jing-Shu Zheng", "Xi Yang", "Yongtao Wang"], "categories": ["cs.RO", "cs.AI"], "id": "2606.16826", "title": "ATOM-Bench: A Real-World Benchmark for Atomic Skills and Compositional Generalization in Manipulation Policies", "type": "arxiv", "url": "https://arxiv.org/abs/2606.16826"}, "sourceUpdatedAt": "2026-06-15T15:08:42Z", "topics": ["Robotics"], "venueAttempts": [], "whyItMatters": "ATOM-Bench provides a public diagnostic testbed for disentangling failures in motor execution, instruction grounding, and compositional reuse, which is crucial for advancing generalist manipulation policies beyond demonstrated tasks."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
