{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2605.30346", "kind": "paper", "url": "https://arxiv.org/abs/2605.30346"}, {"id": "gh:youzhe0305/yocausal", "kind": "repo", "url": "https://github.com/youzhe0305/YoCausal"}, {"kind": "website", "url": "https://www.youzhexie.me/papers/YoCausal/index.html"}], "authors": ["You-Zhe Xie", "Yu-Hsuan Li", "Jie-Ying Lee", "Kaipeng Zhang", "Yu-Lun Liu", "Zhixiang Wang"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2605.30346", "source_url": "https://arxiv.org/abs/2605.30346"}, "categories": ["General AI", "Vision & 3D", "cs.CV"], "description": {"en": "YoCausal is a two-level benchmark that evaluates video diffusion models' understanding of temporal causality using the Violation of Expectation paradigm. It temporally reverses real-world videos as counterfactual samples and introduces two metrics: the Reverse Surprise Index (RSI) for arrow-of-time perception and the Causality Cognition Index (CCI) for disentangling genuine causal reasoning from temporal bias."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2605.30346", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2605.30346", "title": "YoCausal"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2605.30346", "modality": null, "name": "YoCausal", "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_yocausal_0e0be7af", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "ab780266e155cdc0a48dbdaf428be362f51dd20a32c45d0c37f347c6a6a8eb35", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2605.30346"}, "publisher": null, "released": "2026-05-28", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2605.30346", "source_url": "https://arxiv.org/abs/2605.30346"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2605-30346", "source": "claire_radar", "source_benchmark_id": "2605.30346", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Machine Generated", "applicationDomains": ["General AI"], "area": "Vision & 3D", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/youzhe0305/YoCausal", "githubScope": "benchmark_repo", "githubStars": 37, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 52, "hfPaperUrl": "https://huggingface.co/papers/2605.30346", "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", "lastSuccessfulDate": "2026-09-30", "state": "stale"}}, "status": "fresh"}, "capabilities": [], "confidence": "High", "construction": "Original Synthetic", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "c28262a53045ac1b371f3ee4cf2456813b0634cd54fc48006be0075aab8cde90", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "YoCausal is a two-level benchmark that evaluates video diffusion models' understanding of temporal causality using the Violation of Expectation paradigm. It temporally reverses real-world videos as counterfactual samples and introduces two metrics: the Reverse Surprise Index (RSI) for arrow-of-time perception and the Causality Cognition Index (CCI) for disentangling genuine causal reasoning from temporal bias.", "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": "We present YoCausal, a two-level benchmark inspired by the Violation of Expectation (VoE) paradigm from cognitive science."}, "familyId": "bmf_8035539671b8", "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/youzhe0305/YoCausal", "url": "https://github.com/youzhe0305/YoCausal"}, {"basis": "existing-metric-association", "linkedUrl": "https://github.com/youzhe0305/YoCausal", "url": "https://huggingface.co/papers/2605.30346"}, {"basis": "linked-from-project", "linkedUrl": "https://github.com/youzhe0305/YoCausal", "url": "https://www.youzhexie.me/papers/YoCausal/index.html"}], "scope": "benchmark_repo", "url": "https://github.com/youzhe0305/YoCausal", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_yocausal_0e0be7af", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/youzhe0305/YoCausal", "data": null, "hfPaper": "https://huggingface.co/papers/2605.30346", "paper": "https://arxiv.org/abs/2605.30346", "pdf": "https://arxiv.org/pdf/2605.30346", "project": "https://www.youzhexie.me/papers/YoCausal/index.html", "report": "https://arxiv.org/abs/2605.30346"}, "metrics": [], "motivation": "As video diffusion models (VDMs) advance toward world models, a key question arises: do they truly understand causality, or merely overfit to statistical temporal patterns?", "name": "YoCausal", "oneLine": "YoCausal is a two-level benchmark that evaluates video diffusion models' understanding of temporal causality using the Violation of Expectation paradigm. It temporally reverses real-world videos as counterfactual samples and introduces two metrics: the Reverse Surprise Index (RSI) for arrow-of-time perception and the Causality Cognition Index (CCI) for disentangling genuine causal reasoning from temporal bias.", "paperTitle": "YoCausal: How Far is Video Generation from World Model? A Causality Perspective", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2605.30346", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "National Yang Ming Chiao Tung University", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://github.com/youzhe0305/YoCausal"}, {"name": "Shanda AI Research Tokyo", "organizationType": "company-research-lab", "role": "benchmark-publisher", "sourceUrl": "https://github.com/youzhe0305/YoCausal"}], "ranking": {}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-05-28", "source": {"authors": ["You-Zhe Xie", "Yu-Hsuan Li", "Jie-Ying Lee", "Kaipeng Zhang", "Yu-Lun Liu", "Zhixiang Wang"], "categories": ["cs.CV"], "id": "2605.30346", "title": "YoCausal: How Far is Video Generation from World Model? A Causality Perspective", "type": "arxiv", "url": "https://arxiv.org/abs/2605.30346"}, "sourceUpdatedAt": "2026-05-28T17:59:51Z", "topics": ["cs.CV"], "venueAttempts": [], "whyItMatters": "Existing video benchmarks rely on synthetic data and do not separate temporal-direction awareness from true causal understanding. YoCausal provides a protocol that isolates causal cognition from temporal bias, enabling evaluation of whether models genuinely understand cause-and-effect relationships or only statistical patterns."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
