Plug-and-Play Context Feature Reuse for Efficient Masked Generation

TL;DR

ReCAP is a plug-and-play module that accelerates masked generative models by reusing context features, reducing inference time while maintaining high-quality image synthesis.

Contribution

ReCAP introduces a novel feature reuse technique that speeds up masked generative models without sacrificing generation quality.

Findings

ReCAP achieves up to 2.4x faster inference on ImageNet256.

ReCAP maintains high fidelity with reduced computation.

Effective across multiple MGMs and architectures.

Abstract

Masked generative models (MGMs) have emerged as a powerful framework for image synthesis, combining parallel decoding with strong bidirectional context modeling. However, generating high-quality samples typically requires many iterative decoding steps, resulting in high inference costs. A straightforward way to speed up generation is by decoding more tokens in each step, thereby reducing the total number of steps. However, when many tokens are decoded simultaneously, the model can only estimate the univariate marginal distributions independently, failing to capture the dependency among them. As a result, reducing the number of steps significantly compromises generation fidelity. In this work, we introduce ReCAP (Reused Context-Aware Prediction), a plug-and-play module that accelerates inference in MGMs by constructing low-cost steps via reusing feature embeddings from previously decoded context tokens. ReCAP interleaves standard full evaluations with lightweight steps that cache and reuse context features, substantially reducing computation while preserving the benefits of fine-grained, iterative generation. We demonstrate its effectiveness on top of three representative MGMs (MaskGIT, MAGE, and MAR), including both discrete and continuous token spaces and covering diverse architectural designs. In particular, on ImageNet256 class-conditional generation, ReCAP achieves up to 2.4x faster inference than the base model with minimal performance drop, and consistently delivers better efficiency-fidelity trade-offs under various generation settings.