{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.07962", "kind": "paper", "url": "https://arxiv.org/abs/2606.07962"}], "authors": ["Bin Zhu", "Yanhao Jia", "Kexin Zhao", "Jie Wang", "Munan Ning", "Hao Li", "Yuwei Niu", "Tanqing Sun", "Huangchong Yan", "Mingjun Pan", "Xinyi Wu", "Qishen Yin", "Yunyang Ge", "Shuai Zhao", "Li Yuan"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.07962", "source_url": "https://arxiv.org/abs/2606.07962"}, "categories": ["Cybersecurity", "Multimodal", "Reasoning", "Safety & Trustworthiness"], "description": {"en": "ChronoPhyBench evaluates multimodal LLMs on chronological physical dynamics reasoning via next-state prediction and VQA, using video frames and captions for single-image selection and multi-frame sorting."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.07962", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.07962", "title": "ChronoPhyBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Nanyang Technological University", "Peking University", "Peking University Shenzhen Hospital", "Peng Cheng Laboratory", "Tsinghua University"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2606.07962", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2606.07962", "source_url": "https://openalex.org/W7164032484"}, "key": "claire-radar:2606.07962", "modality": null, "name": "ChronoPhyBench", "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_chronophybench_c3f107b4", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "4ec43f48e9408fbbd89d391d88d9143468205e46f9d49fb87c4b070a9dba5de8", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.07962"}, "publisher": null, "released": "2026-06-06", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.07962", "source_url": "https://arxiv.org/abs/2606.07962"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-07962", "source": "claire_radar", "source_benchmark_id": "2606.07962", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["Cybersecurity"], "area": "Safety & Trustworthiness", "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/2606.07962", "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": ["Reasoning"], "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": "5bd1375b6ae9121f7d0eef7ce0cf80d9d83dde07d58b319f355faf630e17b75a", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "ChronoPhyBench evaluates multimodal LLMs on chronological physical dynamics reasoning via next-state prediction and VQA, using video frames and captions for single-image selection and multi-frame sorting.", "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"], "snippet": "Motivated by this, and to rigorously mitigate language modality bias and shortcuts, we propose a novel multimodal Chrono}logical Physical Dynamics Reasoning Benchmark ChronoPhyBench, which unifies next state prediction with Visual Question Answering (VQA) paradigms by conditioning on historical video context and textual captions to enforce models to deduce subsequent physical states through both single image selection and the inherently more complex task of multiple frame chronological sorting."}, "familyId": "bmf_b04565b83dc6", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_chronophybench_c3f107b4", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": ["Cybersecurity"], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2606.07962", "paper": "https://arxiv.org/abs/2606.07962", "pdf": "https://arxiv.org/pdf/2606.07962", "project": null, "report": "https://arxiv.org/abs/2606.07962"}, "metrics": [], "motivation": "Recent advancements in Multimodal Large Language Models (MLLMs) have demonstrated remarkable proficiency in open-world reasoning and understanding.", "name": "ChronoPhyBench", "oneLine": "ChronoPhyBench evaluates multimodal LLMs on chronological physical dynamics reasoning via next-state prediction and VQA, using video frames and captions for single-image selection and multi-frame sorting.", "paperTitle": "ChronoPhyBench: Do MLLMs Truly Understand the World or Merely Exploit Language Priors?", "primaryDomain": "Cybersecurity", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.07962", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "ChronoPhyBench Team", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://arxiv.org/abs/2606.07962"}], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-06-06", "source": {"authors": ["Bin Zhu", "Yanhao Jia", "Kexin Zhao", "Jie Wang", "Munan Ning", "Hao Li", "Yuwei Niu", "Tanqing Sun", "Huangchong Yan", "Mingjun Pan", "Xinyi Wu", "Qishen Yin", "Yunyang Ge", "Shuai Zhao", "Li Yuan"], "categories": ["cs.CV"], "id": "2606.07962", "title": "ChronoPhyBench: Do MLLMs Truly Understand the World or Merely Exploit Language Priors?", "type": "arxiv", "url": "https://arxiv.org/abs/2606.07962"}, "sourceUpdatedAt": "2026-06-06T03:40:47Z", "topics": ["Multimodal", "Reasoning"], "venueAttempts": [], "whyItMatters": "Addresses the gap of distinguishing true multimodal reasoning from language-prior exploitation, providing a robust framework to measure physical reasoning and hallucination rates in models."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
