We propose Next Forcing, a multi-chunk prediction framework that overcomes the myopic<br>supervision problem of autoregressive video world models. Next Forcing establishes new state-of-the-art results on the RoboTwin benchmark (94.1/93.5% on Clean/Random) with significantly faster training convergence.</p>\n","updatedAt":"2026-06-10T13:07:26.404Z","author":{"_id":"6777a782cb3b36883e4c99d7","avatarUrl":"/avatars/b203a55dd12e95621ffabef78d54d33a.svg","fullname":"Gangwei Xu","name":"gangweix","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":7,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8674889802932739},"editors":["gangweix"],"editorAvatarUrls":["/avatars/b203a55dd12e95621ffabef78d54d33a.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.11187","authors":[{"_id":"6a296081887fb79cbf65d685","name":"Gangwei Xu","hidden":false},{"_id":"6a296081887fb79cbf65d686","name":"Qihang Zhang","hidden":false},{"_id":"6a296081887fb79cbf65d687","name":"Jiaming Zhou","hidden":false},{"_id":"6a296081887fb79cbf65d688","name":"Xing Zhu","hidden":false},{"_id":"6a296081887fb79cbf65d689","name":"Yujun Shen","hidden":false},{"_id":"6a296081887fb79cbf65d68a","name":"Xin Yang","hidden":false},{"_id":"6a296081887fb79cbf65d68b","name":"Yinghao Xu","hidden":false}],"publishedAt":"2026-06-09T00:00:00.000Z","submittedOnDailyAt":"2026-06-10T00:00:00.000Z","title":"Next Forcing: Causal World Modeling with Multi-Chunk Prediction","submittedOnDailyBy":{"_id":"6777a782cb3b36883e4c99d7","avatarUrl":"/avatars/b203a55dd12e95621ffabef78d54d33a.svg","isPro":true,"fullname":"Gangwei Xu","user":"gangweix","type":"user","name":"gangweix"},"summary":"Autoregressive video generation has emerged as a powerful paradigm for World Action Models (WAMs). However, existing approaches suffer from slow training convergence and limited converged accuracy, particularly at high frame rates, as the training supervision is confined to the current chunk without explicit signals about future dynamics; they also suffer from slow inference due to iterative video denoising. In this paper, we present Next Forcing, a multi-chunk prediction (MCP) framework for causal world modeling that enables faster training, higher accuracy, and accelerated inference. Inspired by multi-token prediction in large language models, Next Forcing introduces an MCP training objective that augments the main model with lightweight auxiliary MCP modules to simultaneously denoise video chunks at multiple future temporal horizons (next^1, next^2, next^3 chunks). These MCP modules form a causal chain across prediction depths, where intermediate features fused from multiple layers of the main model are leveraged to predict future dynamics, allowing near-future predictions to inform farther-future ones and providing dense multi-scale temporal supervision back to the main model. During training, the MCP modules significantly accelerate convergence and improve converged accuracy, especially at high frame rates: at 50 fps, Next Forcing achieves a 93.1% relative improvement over LingBot-VA at 5k training steps and 2.3x faster convergence, and establishes new state-of-the-art results on the RoboTwin benchmark (94.1/93.5% on Clean/Random). At inference, the MCP modules can be retained to predict the next video chunk in parallel with the current one, achieving 2x inference acceleration. Next Forcing also demonstrates significant improvements on PhyWorld, a benchmark evaluating adherence to physical laws in video generation, and over 50% FVD reduction on general video pretraining.","upvotes":3,"discussionId":"6a296081887fb79cbf65d68c","projectPage":"https://gangweix.github.io/next-forcing/","githubRepo":"https://github.com/gangweix/next-forcing","githubRepoAddedBy":"user","ai_summary":"Next Forcing introduces a multi-chunk prediction framework that accelerates training and inference for autoregressive video generation while improving accuracy and physical law adherence.","ai_keywords":["autoregressive video generation","World Action Models","multi-chunk prediction","causal world modeling","video denoising","temporal horizons","causal chain","multi-scale temporal supervision","inference acceleration","PhyWorld","FVD"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","githubStars":29},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"6777a782cb3b36883e4c99d7","avatarUrl":"/avatars/b203a55dd12e95621ffabef78d54d33a.svg","isPro":true,"fullname":"Gangwei Xu","user":"gangweix","type":"user"},{"_id":"6721acf7bbf9703bc285e840","avatarUrl":"/avatars/487f5cae476758d318465aa9ed103568.svg","isPro":false,"fullname":"Yujie Zhao","user":"HomieZ","type":"user"},{"_id":"6717c5c36bc2876059ed23ab","avatarUrl":"/avatars/52c68fb315760df5ef9323cd8ada5a3c.svg","isPro":false,"fullname":"Xin Zhou","user":"LMD0311","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2606/2606.11187.md"}">
Next Forcing: Causal World Modeling with Multi-Chunk Prediction
Abstract
Next Forcing introduces a multi-chunk prediction framework that accelerates training and inference for autoregressive video generation while improving accuracy and physical law adherence.
Autoregressive video generation has emerged as a powerful paradigm for World Action Models (WAMs). However, existing approaches suffer from slow training convergence and limited converged accuracy, particularly at high frame rates, as the training supervision is confined to the current chunk without explicit signals about future dynamics; they also suffer from slow inference due to iterative video denoising. In this paper, we present Next Forcing, a multi-chunk prediction (MCP) framework for causal world modeling that enables faster training, higher accuracy, and accelerated inference. Inspired by multi-token prediction in large language models, Next Forcing introduces an MCP training objective that augments the main model with lightweight auxiliary MCP modules to simultaneously denoise video chunks at multiple future temporal horizons (next^1, next^2, next^3 chunks). These MCP modules form a causal chain across prediction depths, where intermediate features fused from multiple layers of the main model are leveraged to predict future dynamics, allowing near-future predictions to inform farther-future ones and providing dense multi-scale temporal supervision back to the main model. During training, the MCP modules significantly accelerate convergence and improve converged accuracy, especially at high frame rates: at 50 fps, Next Forcing achieves a 93.1% relative improvement over LingBot-VA at 5k training steps and 2.3x faster convergence, and establishes new state-of-the-art results on the RoboTwin benchmark (94.1/93.5% on Clean/Random). At inference, the MCP modules can be retained to predict the next video chunk in parallel with the current one, achieving 2x inference acceleration. Next Forcing also demonstrates significant improvements on PhyWorld, a benchmark evaluating adherence to physical laws in video generation, and over 50% FVD reduction on general video pretraining.
Community
We propose Next Forcing, a multi-chunk prediction framework that overcomes the myopic
supervision problem of autoregressive video world models. Next Forcing establishes new state-of-the-art results on the RoboTwin benchmark (94.1/93.5% on Clean/Random) with significantly faster training convergence.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2606.11187 in a model README.md to link it from this page.
Cite arxiv.org/abs/2606.11187 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2606.11187 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.