Deprecated: The each() function is deprecated. This message will be suppressed on further calls in /home/zhenxiangba/zhenxiangba.com/public_html/phproxy-improved-master/index.php on line 456 Paper page - VisPhyWorld: Probing Physical Reasoning via Code-Driven Video Reconstruction
https://github.com/TIGER-AI-Lab/VisPhyWorld\n","updatedAt":"2026-02-17T22:32:09.114Z","author":{"_id":"68cccfe16e4618473d571b1f","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/68cccfe16e4618473d571b1f/xeF5Ojc5Q1qvtnWy7XoQL.jpeg","fullname":"Jiarong Liang","name":"lllqaq","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8281652331352234},"editors":["lllqaq"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/68cccfe16e4618473d571b1f/xeF5Ojc5Q1qvtnWy7XoQL.jpeg"],"reactions":[],"isReport":false}},{"id":"699518f2e413ac89df44df02","author":{"_id":"63d3e0e8ff1384ce6c5dd17d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1674830754237-63d3e0e8ff1384ce6c5dd17d.jpeg","fullname":"Librarian Bot (Bot)","name":"librarian-bot","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":318,"isUserFollowing":false},"createdAt":"2026-02-18T01:42:10.000Z","type":"comment","data":{"edited":false,"hidden":false,"latest":{"raw":"This is an automated message from the [Librarian Bot](https://huggingface.co/librarian-bots). I found the following papers similar to this paper. \n\nThe following papers were recommended by the Semantic Scholar API \n\n* [PhysicsMind: Sim and Real Mechanics Benchmarking for Physical Reasoning and Prediction in Foundational VLMs and World Models](https://huggingface.co/papers/2601.16007) (2026)\n* [Code2Worlds: Empowering Coding LLMs for 4D World Generation](https://huggingface.co/papers/2602.11757) (2026)\n* [Thinking with Drafting: Optical Decompression via Logical Reconstruction](https://huggingface.co/papers/2602.11731) (2026)\n* [TimeBlind: A Spatio-Temporal Compositionality Benchmark for Video LLMs](https://huggingface.co/papers/2602.00288) (2026)\n* [TangramPuzzle: Evaluating Multimodal Large Language Models with Compositional Spatial Reasoning](https://huggingface.co/papers/2601.16520) (2026)\n* [Can LLMs See Without Pixels? Benchmarking Spatial Intelligence from Textual Descriptions](https://huggingface.co/papers/2601.03590) (2026)\n* [SimuScene: Training and Benchmarking Code Generation to Simulate Physical Scenarios](https://huggingface.co/papers/2602.10840) (2026)\n\n\n Please give a thumbs up to this comment if you found it helpful!\n\n If you want recommendations for any Paper on Hugging Face checkout [this](https://huggingface.co/spaces/librarian-bots/recommend_similar_papers) Space\n\n You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: `@librarian-bot recommend`","html":"
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
\n
The following papers were recommended by the Semantic Scholar API
Please give a thumbs up to this comment if you found it helpful!
\n
If you want recommendations for any Paper on Hugging Face checkout this Space
\n
You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: \n\n@librarian-bot\n\t recommend
\n","updatedAt":"2026-02-18T01:42:10.544Z","author":{"_id":"63d3e0e8ff1384ce6c5dd17d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1674830754237-63d3e0e8ff1384ce6c5dd17d.jpeg","fullname":"Librarian Bot (Bot)","name":"librarian-bot","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":318,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.7134860157966614},"editors":["librarian-bot"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/1674830754237-63d3e0e8ff1384ce6c5dd17d.jpeg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2602.13294","authors":[{"_id":"6993ffc550fb2c0be4783dcf","user":{"_id":"68cccfe16e4618473d571b1f","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/68cccfe16e4618473d571b1f/xeF5Ojc5Q1qvtnWy7XoQL.jpeg","isPro":false,"fullname":"Jiarong Liang","user":"lllqaq","type":"user"},"name":"Jiarong Liang","status":"claimed_verified","statusLastChangedAt":"2026-02-19T09:53:48.327Z","hidden":false},{"_id":"6993ffc550fb2c0be4783dd0","user":{"_id":"631d760344503b7227837242","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/631d760344503b7227837242/3b6JRusFX6GKJpsN9ZdeJ.png","isPro":false,"fullname":"Max Ku","user":"vinesmsuic","type":"user"},"name":"Max Ku","status":"claimed_verified","statusLastChangedAt":"2026-02-18T09:07:10.293Z","hidden":false},{"_id":"6993ffc550fb2c0be4783dd1","name":"Ka-Hei Hui","hidden":false},{"_id":"6993ffc550fb2c0be4783dd2","name":"Ping Nie","hidden":false},{"_id":"6993ffc550fb2c0be4783dd3","name":"Wenhu Chen","hidden":false}],"publishedAt":"2026-02-09T05:46:44.000Z","submittedOnDailyAt":"2026-02-17T20:02:09.103Z","title":"VisPhyWorld: Probing Physical Reasoning via Code-Driven Video Reconstruction","submittedOnDailyBy":{"_id":"68cccfe16e4618473d571b1f","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/68cccfe16e4618473d571b1f/xeF5Ojc5Q1qvtnWy7XoQL.jpeg","isPro":false,"fullname":"Jiarong Liang","user":"lllqaq","type":"user"},"summary":"Evaluating whether Multimodal Large Language Models (MLLMs) genuinely reason about physical dynamics remains challenging. Most existing benchmarks rely on recognition-style protocols such as Visual Question Answering (VQA) and Violation of Expectation (VoE), which can often be answered without committing to an explicit, testable physical hypothesis. We propose VisPhyWorld, an execution-based framework that evaluates physical reasoning by requiring models to generate executable simulator code from visual observations. By producing runnable code, the inferred world representation is directly inspectable, editable, and falsifiable. This separates physical reasoning from rendering. Building on this framework, we introduce VisPhyBench, comprising 209 evaluation scenes derived from 108 physical templates and a systematic protocol that evaluates how well models reconstruct appearance and reproduce physically plausible motion. Our pipeline produces valid reconstructed videos in 97.7% on the benchmark. Experiments show that while state-of-the-art MLLMs achieve strong semantic scene understanding, they struggle to accurately infer physical parameters and to simulate consistent physical dynamics.","upvotes":12,"discussionId":"6993ffc550fb2c0be4783dd4","projectPage":"https://tiger-ai-lab.github.io/VisPhyWorld/","githubRepo":"https://github.com/TIGER-AI-Lab/VisPhyWorld","githubRepoAddedBy":"user","ai_summary":"VisPhyWorld framework evaluates physical reasoning in MLLMs by requiring executable simulator code generation from visual observations, separating physical reasoning from rendering and enabling inspectable, falsifiable world representations.","ai_keywords":["Multimodal Large Language Models","Visual Question Answering","Violation of Expectation","execution-based framework","simulator code generation","physical reasoning","world representation","physical parameters","physical dynamics","semantic scene understanding"],"githubStars":3,"organization":{"_id":"6313a90017838d05194fd282","name":"TIGER-Lab","fullname":"TIGER-Lab","avatar":"https://cdn-uploads.huggingface.co/production/uploads/6313a86154e6e5d9f0f94e04/Noi3Qq3RYz8Jdq6BaFteq.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"631d760344503b7227837242","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/631d760344503b7227837242/3b6JRusFX6GKJpsN9ZdeJ.png","isPro":false,"fullname":"Max Ku","user":"vinesmsuic","type":"user"},{"_id":"63041b541dd5d3c62486c294","avatarUrl":"/avatars/a5286d562f7b9082730f760e66c3bf29.svg","isPro":true,"fullname":"Ho Kei Cheng","user":"hkchengrex","type":"user"},{"_id":"67cd30337fbbaaf96868705f","avatarUrl":"/avatars/380146836cf6ad89f082e62f6c630230.svg","isPro":false,"fullname":"Eddie Tsai","user":"EddieTsai123","type":"user"},{"_id":"6365d5baa7a1324ccd5ecdb9","avatarUrl":"/avatars/636d3f410b878e451a878a6cf171dd53.svg","isPro":false,"fullname":"Thomas Chong","user":"chongcht","type":"user"},{"_id":"64f8e358766ff9f3d2b0de84","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64f8e358766ff9f3d2b0de84/R2P1YG-mRBh7TU9wkjGGk.jpeg","isPro":true,"fullname":"Cong Wei","user":"CongWei1230","type":"user"},{"_id":"670d7aed372cb8fadbd270bb","avatarUrl":"/avatars/652c8b90d492e3e5db6c735d69ae0991.svg","isPro":false,"fullname":"Songcheng Cai","user":"SongchengCai","type":"user"},{"_id":"64405a9d518271b0d1beea38","avatarUrl":"/avatars/b702474588fd7090773320422417a582.svg","isPro":false,"fullname":"Weiming Ren","user":"wren93","type":"user"},{"_id":"68cccfe16e4618473d571b1f","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/68cccfe16e4618473d571b1f/xeF5Ojc5Q1qvtnWy7XoQL.jpeg","isPro":false,"fullname":"Jiarong Liang","user":"lllqaq","type":"user"},{"_id":"684c8b7349cb60c8c190d0a0","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/684c8b7349cb60c8c190d0a0/8916yYbA4GrAI065DWoow.jpeg","isPro":false,"fullname":"Xiangchao Chen","user":"UlyssesXC","type":"user"},{"_id":"6313a86154e6e5d9f0f94e04","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1662232951344-6313a86154e6e5d9f0f94e04.jpeg","isPro":false,"fullname":"Wenhu Chen","user":"wenhu","type":"user"},{"_id":"66bf00ca5b4e241fe266059d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/66bf00ca5b4e241fe266059d/VoWPC_C4zoeT6dS699t7L.png","isPro":false,"fullname":"Keming Wu","user":"wukeming11","type":"user"},{"_id":"64de37ee5e192985054be575","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64de37ee5e192985054be575/fVV7JQMtp_J3uFqszJJHH.jpeg","isPro":false,"fullname":"Yuansheng Ni","user":"yuanshengni","type":"user"}],"acceptLanguages":["*"],"dailyPaperRank":0,"organization":{"_id":"6313a90017838d05194fd282","name":"TIGER-Lab","fullname":"TIGER-Lab","avatar":"https://cdn-uploads.huggingface.co/production/uploads/6313a86154e6e5d9f0f94e04/Noi3Qq3RYz8Jdq6BaFteq.png"}}">
VisPhyWorld framework evaluates physical reasoning in MLLMs by requiring executable simulator code generation from visual observations, separating physical reasoning from rendering and enabling inspectable, falsifiable world representations.
AI-generated summary
Evaluating whether Multimodal Large Language Models (MLLMs) genuinely reason about physical dynamics remains challenging. Most existing benchmarks rely on recognition-style protocols such as Visual Question Answering (VQA) and Violation of Expectation (VoE), which can often be answered without committing to an explicit, testable physical hypothesis. We propose VisPhyWorld, an execution-based framework that evaluates physical reasoning by requiring models to generate executable simulator code from visual observations. By producing runnable code, the inferred world representation is directly inspectable, editable, and falsifiable. This separates physical reasoning from rendering. Building on this framework, we introduce VisPhyBench, comprising 209 evaluation scenes derived from 108 physical templates and a systematic protocol that evaluates how well models reconstruct appearance and reproduce physically plausible motion. Our pipeline produces valid reconstructed videos in 97.7% on the benchmark. Experiments show that while state-of-the-art MLLMs achieve strong semantic scene understanding, they struggle to accurately infer physical parameters and to simulate consistent physical dynamics.
Evaluating whether Multimodal Large Language Models (MLLMs) genuinely reason about physical dynamics remains challenging. Most existing benchmarks rely on recognition-style protocols such as Visual Question Answering (VQA) and Violation of Expectation (VoE), which can often be answered without committing to an explicit, testable physical hypothesis. We propose VisPhyWorld, an execution-based framework that evaluates physical reasoning by requiring models to generate executable simulator code from visual observations. By producing runnable code, the inferred world representation is directly inspectable, editable, and falsifiable. This separates physical reasoning from rendering. Building on this framework, we introduce VisPhyBench, comprising 209 evaluation scenes derived from 108 physical templates and a systematic protocol that evaluates how well models reconstruct appearance and reproduce physically plausible motion. Our pipeline produces valid reconstructed videos in 97.7% on the benchmark. Experiments show that while state-of-the-art MLLMs achieve strong semantic scene understanding, they struggle to accurately infer physical parameters and to simulate consistent physical dynamics.