WorldCache: Accelerating World Models for Free via Heterogeneous Token Caching

• URL: http://arxiv.org/abs/2603.06331v1
• Date: Fri, 06 Mar 2026 14:43:47 GMT
• Title: WorldCache: Accelerating World Models for Free via Heterogeneous Token Caching
• Abstract summary: We propose textbfWorldCache, a caching framework tailored to diffusion world models.<n>We show that WorldCache delivers up to textbf3.7$times$ end-to-end speedups while maintaining textbf98% rollout quality.
• Score: 50.43356949257532
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Diffusion-based world models have shown strong potential for unified world simulation, but the iterative denoising remains too costly for interactive use and long-horizon rollouts. While feature caching can accelerate inference without training, we find that policies designed for single-modal diffusion transfer poorly to world models due to two world-model-specific obstacles: \emph{token heterogeneity} from multi-modal coupling and spatial variation, and \emph{non-uniform temporal dynamics} where a small set of hard tokens drives error growth, making uniform skipping either unstable or overly conservative. We propose \textbf{WorldCache}, a caching framework tailored to diffusion world models. We introduce \textit{Curvature-guided Heterogeneous Token Prediction}, which uses a physics-grounded curvature score to estimate token predictability and applies a Hermite-guided damped predictor for chaotic tokens with abrupt direction changes. We also design \textit{Chaotic-prioritized Adaptive Skipping}, which accumulates a curvature-normalized, dimensionless drift signal and recomputes only when bottleneck tokens begin to drift. Experiments on diffusion world models show that WorldCache delivers up to \textbf{3.7$\times$} end-to-end speedups while maintaining \textbf{98\%} rollout quality, demonstrating the vast advantages and practicality of WorldCache in resource-constrained scenarios. Our code is released in https://github.com/FofGofx/WorldCache.

Related papers

• From Local Mismatch to Global Impact: Optimizing Cache Reuse Policy for Efficient Diffusion [29.046432574690602]
  Cache-based acceleration has emerged as a promising solution, but existing policies on local similaritys are misaligned with final generation quality.<n>This discrepancy stems from the non-uniform propagation and accumulation of errors along the denoising trajectory.<n>We propose Global Cache Cache (GCache) to address this problem.<n>GCache consistently outperforms prior caching strategies for both video and image generation.
  arXiv  Detail & Related papers  (2026-08-13T10:08:47Z)
• OnlineCache: Learning Dynamic Caching Policies with Error Correction for Efficient Diffusion Inference [21.088549125949548]
  Diffusion models have revolutionized generative tasks but incur high latency due to iterative denoising.<n>OnlineCache is a dynamic caching framework that jointly learns when to cache and how to correct approximation errors.<n>Our method automatically allocates computational resources across both samples and timesteps, improving overall generation quality.
  arXiv  Detail & Related papers  (2026-07-31T13:17:05Z)
• Motion-Aware Caching for Efficient Autoregressive Video Generation [73.27762884730272]
  MotionCache exploits inter-frame differences as a lightweight proxy for pixel-level motion characteristics.<n>Experiments on state-of-the-art models like SkyReels-V2 and MAGI-1 demonstrate that MotionCache achieves significant speedups.
  arXiv  Detail & Related papers  (2026-05-03T05:49:27Z)
• WorldCache: Content-Aware Caching for Accelerated Video World Models [50.7543797435026]
  We present textbfWorldCache, a Perception-Constrained Dynamic CaCache framework.<n>WorldCache improves both when and how to reuse features.<n>On Cosmos-2.5-2B evaluated on PAI-Bench, WorldCache achieves textbf$2.3times$ inference speedup while preserving textbf99.4% of baseline quality.
  arXiv  Detail & Related papers  (2026-03-23T17:59:54Z)
• SenCache: Accelerating Diffusion Model Inference via Sensitivity-Aware Caching [75.02865981328509]
  Caching reduces computation by reusing previously computed model outputs across timesteps.<n>We propose Sensitivity-Aware Caching (SenCache), a dynamic caching policy that adaptively selects caching timesteps on a per-sample basis.<n>SenCache achieves better visual quality than existing caching methods under similar computational budgets.
  arXiv  Detail & Related papers  (2026-02-27T17:36:09Z)
• Denoising as Path Planning: Training-Free Acceleration of Diffusion Models with DPCache [8.614492355393578]
  We propose DPCache, a training-free acceleration framework that formulates diffusion acceleration as a global path planning problem.<n> DPCache employs dynamic programming to select an optimal sequence of key timesteps that minimizes the total path cost while preserving trajectory fidelity.<n>Experiments on DiT, FLUX, and HunyuanVideo demonstrate that DPCache achieves strong acceleration with minimal quality loss.
  arXiv  Detail & Related papers  (2026-02-26T06:13:33Z)
• Fast-SAM3D: 3Dfy Anything in Images but Faster [65.17322167628367]
  SAM3D enables scalable, open-world 3D reconstruction from complex scenes, yet its deployment is hindered by prohibitive inference latency.<n>We present textbfFast-SAM3D, a training-free framework that aligns computation with instantaneous generation complexity.
  arXiv  Detail & Related papers  (2026-02-05T04:27:59Z)
• DiCache: Let Diffusion Model Determine Its Own Cache [62.954717254728166]
  DiCache is a training-free adaptive caching strategy for accelerating diffusion models at runtime.<n>Online Probe Profiling Scheme leverages a shallow-layer online probe to obtain an on-the-fly indicator for the caching error in real time.<n> Dynamic Cache Trajectory Alignment approximates the deep-layer feature output from multi-step historical caches.
  arXiv  Detail & Related papers  (2025-08-24T13:30:00Z)
• SlimInfer: Accelerating Long-Context LLM Inference via Dynamic Token Pruning [3.502168555273189]
  SlimInfer aims to accelerate inference by directly pruning less critical prompt tokens during the forward pass.<n>We show that SlimInfer can achieve up to $mathbf2.53times$ time-to-first-token (TTFT) speedup and $mathbf1.88times$ end-to-end latency reduction for LLaMA3.1-8B-Instruct.
  arXiv  Detail & Related papers  (2025-08-08T16:42:38Z)
• Compressed and Smooth Latent Space for Text Diffusion Modeling [71.87805084454187]
  We introduce Cosmos, a novel approach to text generation that operates entirely in a compressed, smooth latent space tailored specifically for diffusion.<n>We demonstrate that text representations can be compressed by $8times$ while maintaining generation quality comparable to token-level diffusion models.<n>We evaluate Cosmos on four diverse generative tasks including story generation, question generation, summarization, and detoxification and compare it with various generative paradigms.
  arXiv  Detail & Related papers  (2025-06-26T12:05:13Z)
• Model Reveals What to Cache: Profiling-Based Feature Reuse for Video Diffusion Models [41.11005178050448]
  ProfilingDiT is a novel adaptive caching strategy that explicitly disentangles foreground and background-focused blocks.<n>Our framework achieves significant acceleration while maintaining visual fidelity across comprehensive quality metrics.
  arXiv  Detail & Related papers  (2025-04-04T03:30:15Z)
• DeepCache: Accelerating Diffusion Models for Free [65.02607075556742]
  DeepCache is a training-free paradigm that accelerates diffusion models from the perspective of model architecture. DeepCache capitalizes on the inherent temporal redundancy observed in the sequential denoising steps of diffusion models. Under the same throughput, DeepCache effectively achieves comparable or even marginally improved results with DDIM or PLMS.
  arXiv  Detail & Related papers  (2023-12-01T17:01:06Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.