{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.24056", "kind": "paper", "url": "https://arxiv.org/abs/2608.24056"}, {"id": "gh:narnialabs/leaderboard", "kind": "repo", "url": "https://github.com/Narnialabs/leaderboard"}, {"kind": "website", "url": "https://leaderboard.narnia.ai"}], "authors": ["Sang Won Lee", "Hyogu Jeong", "Namwoo Kang"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.24056", "source_url": "https://arxiv.org/abs/2608.24056"}, "categories": ["General AI", "Language & Knowledge", "cs.LG"], "description": {"en": "PhysicsBench is a unified benchmark and leaderboard for generative and predictive AI models in engineering design and simulation. It spans seven tasks across 1D, 2D, and 3D domains, ranks 66 models on nine datasets, and uses a common metric suite including BenchRank for debiased ranking. Evaluation covers data scales from S to XL."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.24056", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.24056", "title": "PhysicsBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Korea Advanced Institute of Science and Technology"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2608.24056", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2608.24056", "source_url": "https://openalex.org/W7204266034"}, "key": "claire-radar:2608.24056", "modality": null, "name": "PhysicsBench", "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": "true", "origin_record_id": "bm_physicsbench_c13eff48", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "13dbc63e87c5495f84e4842e05ddb491afdfc96ba878abe39f20c34206e12d1a", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-08-27T04:18:56.318436Z", "source_url": "https://arxiv.org/abs/2608.24056"}, "publisher": null, "released": "2026-08-25", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.24056", "source_url": "https://arxiv.org/abs/2608.24056"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-24056", "source": "claire_radar", "source_benchmark_id": "2608.24056", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Language & Knowledge", "attention": {"asOf": "2026-08-26", "attemptedAt": "2026-08-26T11:32:18.902760Z", "githubRepo": "https://github.com/Narnialabs/leaderboard", "githubScope": "benchmark_repo", "githubStars": 0, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2608.24056", "observedAt": "2026-08-26T11:32:18.902760Z", "signalStatus": {"githubStars": {"observedAt": "2026-08-26T11:32:18.902760Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-08-26T11:32:18.902760Z", "state": "unavailable"}}, "status": "fresh"}, "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "Practical engineering relevance and a live leaderboard with broad model coverage are likely to draw industrial and academic attention.", "score": 80}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-27T04:22:25.189955Z", "inputHash": "13a6a5fd5b74275848082aa511047279b8707786f76252cc1bbf8f25a1d922ef", "model": "deepseek-v4-flash"}, "curation": {"canonicalNameEvidence": "We present PhysicsBench, a unified benchmark and leaderboard", "canonicalNameSource": "abstract", "decisionReason": "The benchmark has an official project page, live leaderboard, and code repository, indicating a public reuse path and stable scoring contract.", "model": "deepseek-v4-pro", "reviewedAt": "2026-08-27T04:18:56.318436Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "PhysicsBench is a unified benchmark and leaderboard for generative and predictive AI models in engineering design and simulation. It spans seven tasks across 1D, 2D, and 3D domains, ranks 66 models on nine datasets, and uses a common metric suite including BenchRank for debiased ranking. Evaluation covers data scales from S to XL.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "score_submission", "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "We present PhysicsBench, a unified benchmark and leaderboard that evaluates generative and predictive models under one standardized procedure."}, "familyId": "bmf_b34dce60cd36", "firstSeenAt": "2026-08-26", "heat": null, "id": "bm_physicsbench_c13eff48", "indexedAt": "2026-08-26T06:08:22.244942Z", "industrySectors": [], "links": {"code": "https://github.com/Narnialabs/leaderboard", "data": null, "paper": "https://arxiv.org/abs/2608.24056", "pdf": "https://arxiv.org/pdf/2608.24056", "project": "https://leaderboard.narnia.ai", "report": "https://arxiv.org/abs/2608.24056"}, "metrics": [], "motivation": "Generative and predictive artificial intelligence models are increasingly used to generate geometry and to predict physical fields and scalar quantities in engineering design and simulation.", "name": "PhysicsBench", "oneLine": "PhysicsBench is a unified benchmark and leaderboard for generative and predictive AI models in engineering design and simulation. It spans seven tasks across 1D, 2D, and 3D domains, ranks 66 models on nine datasets, and uses a common metric suite including BenchRank for debiased ranking. Evaluation covers data scales from S to XL.", "paperTitle": "PhysicsBench: A Unified Leaderboard for Generative and Predictive Models in Engineering Design and Simulation", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.24056", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-26T06:08:22.244942Z"}, "publications": [], "publishers": [{"name": "Narnia Labs", "organizationType": "company-research-lab", "role": "benchmark-publisher", "sourceUrl": "https://leaderboard.narnia.ai"}], "ranking": {"30d": {"confidence": "Low", "coverage": 0.55, "growth": {"components": {"githubStars": {"percentile": null, "value": null}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 30}, "level": {"components": {"githubStars": {"percentile": 0.19387755102040816, "value": 0}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 106, "score": 33}, "method": "attention-ranking-v12", "rank": 106, "score": 33}, "90d": {"confidence": "Low", "coverage": 0.55, "growth": {"components": {"githubStars": {"percentile": null, "value": null}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"percentile": 0.12032085561497326, "value": 0}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 320, "score": 29}, "method": "attention-ranking-v12", "rank": 320, "score": 29}, "today": {"confidence": "Low", "coverage": 0.25, "growth": {"components": {"githubStars": {"percentile": null, "value": null}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": 2, "historyDate": "2026-08-24", "rank": null, "score": null, "windowDays": 1}, "level": {"components": {"githubStars": {"percentile": 0.5, "value": 0}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}, "llmAttentionForecast": {"maxContribution": 0.25, "percentile": 0.73, "role": "bonus", "value": 73}}, "confidence": "Low", "coverage": 0.25, "rank": 1, "score": 56}, "method": "attention-ranking-v12", "rank": 1, "score": 56}}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-08-25", "source": {"authors": ["Sang Won Lee", "Hyogu Jeong", "Namwoo Kang"], "categories": ["cs.LG", "cs.CE"], "id": "2608.24056", "title": "PhysicsBench: A Unified Leaderboard for Generative and Predictive Models in Engineering Design and Simulation", "type": "arxiv", "url": "https://arxiv.org/abs/2608.24056"}, "sourceUpdatedAt": "2026-08-25T04:32:18Z", "topics": ["cs.LG"], "venueAttempts": [], "whyItMatters": "Existing evaluations of generative and predictive AI in engineering are isolated with inconsistent metrics. PhysicsBench provides a standardized leaderboard for model selection, revealing that academic performance poorly predicts small-data ranking, supporting data-efficiency decisions."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
