Next Forcing: Causal World Modeling with Multi-Chunk Prediction
2026-06-09 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors propose Next Forcing, a new method to improve autoregressive video generation by predicting several future video chunks at once instead of just the next one. This approach uses extra helper modules that work together to give the main model better hints about future video frames, which makes training faster and more accurate, especially for videos with high frame rates. Their method also speeds up video generation during use by predicting multiple chunks in parallel. They tested Next Forcing on different benchmarks and showed it outperforms previous methods both in accuracy and speed.
autoregressive video generationworld action modelsmulti-chunk predictionmulti-token predictionvideo denoisingtraining convergenceRoboTwin benchmarkPhyWorld benchmarkframe rateFVD (Fréchet Video Distance)
Authors
Gangwei Xu, Qihang Zhang, Jiaming Zhou, Xing Zhu, Yujun Shen, Xin Yang, Yinghao Xu
Abstract
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.