X-Cache: Cross-Chunk Block Caching for Few-Step Autoregressive World Models Inference

2026-04-22

AI Summary

This work addresses the challenge of deploying few-step autoregressive world models in real-time interactive settings due to their high computational overhead. The authors propose a training-free inference acceleration method that, for the first time, enables block-level residual caching across consecutively generated blocks. A novel structure- and action-aware fingerprint-based dual-gating mechanism dynamically determines whether to reuse cached representations, while a KV-update block identification scheme mitigates error propagation, making the approach suitable for closed-loop interaction scenarios. Evaluated on the X-world model, the method achieves a 71% block-skipping rate and a 2.6× end-to-end speedup with negligible degradation in generation quality.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Diffusion Models for Vision
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Real-time world simulation is becoming a key infrastructure for scalable evaluation and online reinforcement learning of autonomous driving systems. Recent driving world models built on autoregressive video diffusion achieve high-fidelity, controllable multi-camera generation, but their inference cost remains a bottleneck for interactive deployment. However, existing diffusion caching methods are designed for offline video generation with multiple denoising steps, and do not transfer to this scenario. Few-step distilled models have no inter-step redundancy left for these methods to reuse, and sequence-level parallelization techniques require future conditioning that closed-loop interactive generation does not provide. We present X-Cache, a training-free acceleration method that caches along a different axis: across consecutive generation chunks rather than across denoising steps. X-Cache maintains per-block residual caches that persist across chunks, and applies a dual-metric gating mechanism over a structure- and action-aware block-input fingerprint to independently decide whether each block should recompute or reuse its cached residual. To prevent approximation errors from permanently contaminating the autoregressive KV cache, X-Cache identifies KV update chunks (the forward passes that write clean keys and values into the persistent cache) and unconditionally forces full computation on these chunks, cutting off error propagation. We implement X-Cache on X-world, a production multi-camera action-conditioned driving world model built on multi-block causal DiT with few-step denoising and rolling KV cache. X-Cache achieves 71% block skip rate with 2.6x wall-clock speedup while maintaining minimum degradation.

Problem

• autoregressive world models
• few-step diffusion
• real-time simulation
• interactive generation
• inference acceleration

Innovation

• cross-chunk caching
• autoregressive world models
• training-free acceleration
• KV cache protection
• few-step diffusion