{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.27345", "kind": "paper", "url": "https://arxiv.org/abs/2608.27345"}], "authors": ["Yuandong Pu", "Le Zhuo", "Sayak Paul", "Gabriel Jorge Menezes", "Avram Đorđević", "Shiyang Li", "Yifan Zhou", "Bin Fu", "Wenlong Zhang", "Junjun He", "Yu Qiao", "Yihao Liu", "Jinbo Xing", "Xi Chen"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.27345", "source_url": "https://arxiv.org/abs/2608.27345"}, "categories": ["General AI", "Safety", "Vision & 3D"], "description": {"en": "Evaluates video generators as stochastic samplers of world dynamics across 50 scenarios, comparing the distribution of possible behaviors under identical initial observations and actions."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.27345", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.27345", "title": "PAWBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.27345", "modality": null, "name": "PAWBench", "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": "true", "origin_record_id": "bm_pawbench_0877eb45", "origin_source": "arxiv", "recognition_confidence": "0.5", "record_sha256": "f31601322e66ab542599e23b605078d426f0c62827ab90d02bfae0d1f4e8adac", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-08-31T07:23:36.749451Z", "source_url": "https://arxiv.org/abs/2608.27345"}, "publisher": null, "released": "2026-08-27", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.27345", "source_url": "https://arxiv.org/abs/2608.27345"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-27345", "source": "claire_radar", "source_benchmark_id": "2608.27345", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Vision & 3D", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/Andrew0613/PAWBench", "githubScope": "benchmark_repo", "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 75, "hfPaperUrl": "https://huggingface.co/papers/2608.27345", "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": "2026-09-28", "state": "stale"}}, "status": "stale"}, "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "Timely focus on distributional evaluation of world models and release announcement could draw strong interest from the video generation community.", "score": 70}, "capabilities": [], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-31T07:23:36.749451Z", "inputHash": "8690b52f9124fb32dd2a5fece1abe1edd849d60c3ec784c93b19b5f78ba99479", "model": "deepseek-v4-pro"}, "curation": {"canonicalNameEvidence": "introduce PAWBench, a benchmark for evaluating video generators as stochastic samplers of world dynamics", "canonicalNameSource": "abstract", "decisionReason": "Formally named benchmark with explicit evaluation protocol and scoring contract; no public artifact link provided yet, but code and data are indicated for release upon acceptance.", "model": "deepseek-v4-pro", "reviewedAt": "2026-08-31T07:23:36.749451Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Evaluates video generators as stochastic samplers of world dynamics across 50 scenarios, comparing the distribution of possible behaviors under identical initial observations and actions.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark"], "snippet": "To answer it, we formalize probabilistic alignment as a distributional criterion for world models and introduce PAWBench, a benchmark for evaluating video generators as stochastic samplers of world dynamics."}, "familyId": "bmf_d1e745e2c7d4", "firstSeenAt": "2026-08-30", "heat": null, "id": "bm_pawbench_0877eb45", "indexedAt": "2026-08-30T06:59:55.242506Z", "industrySectors": [], "links": {"code": null, "data": null, "paper": "https://arxiv.org/abs/2608.27345", "pdf": "https://arxiv.org/pdf/2608.27345", "project": null, "report": "https://arxiv.org/abs/2608.27345"}, "metrics": [], "motivation": "Recent video generation models are increasingly framed as world models.", "name": "PAWBench", "oneLine": "Evaluates video generators as stochastic samplers of world dynamics across 50 scenarios, comparing the distribution of possible behaviors under identical initial observations and actions.", "paperTitle": "PAWBench: How Far Are We from Probabilistically Aligned World Modeling?", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.27345", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-30T06:59:55.242506Z"}, "publications": [], "ranking": {"90d": {"confidence": "Low", "coverage": 0.15, "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": null, "value": null}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.7652172363448368, "value": 75}}, "confidence": "Low", "coverage": 0.15, "rank": null, "score": 54}, "method": "attention-ranking-v13", "rank": null, "score": 54}}, "readiness": "Paper only", "recognitionConfidence": 0.5, "relation": "introduces", "releasedAt": "2026-08-27", "source": {"authors": ["Yuandong Pu", "Le Zhuo", "Sayak Paul", "Gabriel Jorge Menezes", "Avram {\\DJ}or{\\dj}evi\\'c", "Shiyang Li", "Yifan Zhou", "Bin Fu", "Wenlong Zhang", "Junjun He", "Yu Qiao", "Yihao Liu", "Jingbo Xing", "Xi Chen"], "categories": ["cs.CV", "cs.AI"], "id": "2608.27345", "title": "PAWBench: How Far Are We from Probabilistically Aligned World Modeling?", "type": "arxiv", "url": "https://arxiv.org/abs/2608.27345"}, "sourceUpdatedAt": "2026-08-27T16:46:38Z", "topics": ["Safety"], "venueAttempts": [], "whyItMatters": "Current evaluations assess single-video plausibility, missing whether models recover the correct distribution of physical behavior, a requirement for reliable world modeling."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
