{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2607.25641", "kind": "paper", "url": "https://arxiv.org/abs/2607.25641"}, {"id": "gh:zjukg/omniphys", "kind": "repo", "url": "https://github.com/zjukg/OmniPhys"}], "authors": ["Yajing Xu", "Yarong Lan", "Jiaoyan Chen", "Yichi Zhang", "Jeff Z. Pan", "Mingchen Tu", "Zhizhen Liu", "Wen Zhang", "Huajun Chen"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2607.25641", "source_url": "https://arxiv.org/abs/2607.25641"}, "categories": ["Factuality", "General AI", "Vision & 3D", "cs.CV"], "description": {"en": "OmniPhys is a benchmark of 1,551 text-to-image generation samples grounded in a Physical Knowledge Graph, aligned with PhET simulations and curricula. It evaluates physical commonsense in generated images using a dual-path verification protocol."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2607.25641", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2607.25641", "title": "OmniPhys"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2607.25641", "modality": null, "name": "OmniPhys", "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_omniphys_eb0844b7", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "bbf002a587a022c61454e331c11f260eb3bf65c416c0830e1fb828c618ccb5b2", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2607.25641"}, "publisher": null, "released": "2026-07-28", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2607.25641", "source_url": "https://arxiv.org/abs/2607.25641"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2607-25641", "source": "claire_radar", "source_benchmark_id": "2607.25641", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Mixed", "applicationDomains": ["General AI"], "area": "Vision & 3D", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/zjukg/OmniPhys", "githubScope": "benchmark_repo", "githubStars": 1, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2607.25641", "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": {"checkedAt": "2026-10-04T15:31:48.030916Z", "state": "unavailable"}}, "status": "fresh"}, "capabilities": ["Factuality"], "confidence": "High", "construction": "Aggregate Existing", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "97c19e30b9cdcf993a1194adeb7f5614e69344cc3e4a4a4ee494d460edd5a4c9", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "OmniPhys is a benchmark of 1,551 text-to-image generation samples grounded in a Physical Knowledge Graph, aligned with PhET simulations and curricula. It evaluates physical commonsense in generated images using a dual-path verification protocol.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "To address these challenges, we introduce OmniPhys, a rigorous benchmark of 1,551 samples grounded in a Physical Knowledge Graph."}, "familyId": "bmf_1f087830f505", "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/zjukg/OmniPhys", "url": "https://github.com/zjukg/OmniPhys"}, {"basis": "existing-metric-association", "linkedUrl": "https://github.com/zjukg/OmniPhys", "url": "https://huggingface.co/papers/2607.25641"}, {"basis": "linked-from-report", "linkedUrl": "https://github.com/zjukg/OmniPhys", "url": "https://arxiv.org/abs/2607.25641"}], "scope": "benchmark_repo", "url": "https://github.com/zjukg/OmniPhys", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_omniphys_eb0844b7", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/zjukg/OmniPhys", "data": null, "hfPaper": "https://huggingface.co/papers/2607.25641", "paper": "https://arxiv.org/abs/2607.25641", "pdf": "https://arxiv.org/pdf/2607.25641", "project": null, "report": "https://arxiv.org/abs/2607.25641"}, "metrics": [], "motivation": "While text-to-image models exhibit remarkable visual fidelity, they frequently violate fundamental physical commonsense.", "name": "OmniPhys", "oneLine": "OmniPhys is a benchmark of 1,551 text-to-image generation samples grounded in a Physical Knowledge Graph, aligned with PhET simulations and curricula. It evaluates physical commonsense in generated images using a dual-path verification protocol.", "paperTitle": "OmniPhys: Knowledge-Graph-Driven Benchmarking and Collective Optimization for Physical Commonsense in Text-to-Image Generation", "primaryDomain": "General AI", "publication": {"evidence": "accepted by KDD 2026 DB track", "evidenceLevel": "author-claim", "evidenceUrl": "https://arxiv.org/abs/2607.25641", "source": "arxiv-comments", "status": "acceptance_claimed", "venue": "KDD 2026 DB track", "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "ZJUKG", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://github.com/zjukg/OmniPhys"}], "ranking": {"90d": {"confidence": "Low", "coverage": 0.55, "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": 0.09252268850709358, "value": 1}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 345, "score": 28}, "method": "attention-ranking-v13", "rank": 345, "score": 28}}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-07-28", "source": {"authors": ["Yajing Xu", "Yarong Lan", "Jiaoyan Chen", "Yichi Zhang", "Jeff Z. Pan", "Mingchen Tu", "Zhizhen Liu", "Wen Zhang", "Huajun Chen"], "categories": ["cs.CV", "cs.AI"], "id": "2607.25641", "title": "OmniPhys: Knowledge-Graph-Driven Benchmarking and Collective Optimization for Physical Commonsense in Text-to-Image Generation", "type": "arxiv", "url": "https://arxiv.org/abs/2607.25641"}, "sourceUpdatedAt": "2026-07-28T12:27:25Z", "topics": ["cs.CV"], "venueAttempts": [{"decisionRaw": "accepted by KDD 2026 DB track", "evidence": [{"level": "author-claim", "observedAt": "2026-10-14T01:46:58.058199Z", "rawValue": "accepted by KDD 2026 DB track", "sourceType": "arxiv-comments", "sourceUrl": "https://arxiv.org/abs/2607.25641"}], "reviewStatus": "accepted", "venueName": "KDD 2026 DB track"}], "whyItMatters": "Existing benchmarks use coarse descriptions and fail to diagnose specific physical principles. OmniPhys provides a fine-grained, curriculum-aligned evaluation to identify systemic physical reasoning gaps in image generation models, supporting targeted improvement."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
