SJD-PV: Speculative Jacobi Decoding with Phrase Verification for Autoregressive Image Generation

TL;DR

This paper introduces SJD-PV, a training-free method that accelerates autoregressive image generation by verifying groups of co-occurring tokens as phrases, reducing inference time while maintaining quality.

Contribution

The paper proposes a novel phrase-level speculative verification framework that leverages token co-occurrence statistics for faster autoregressive image decoding without retraining.

Findings

Achieves up to 30% faster decoding speed.

Reduces the number of function evaluations significantly.

Maintains high visual fidelity during accelerated generation.

Abstract

Autoregressive (AR) image models have recently demonstrated remarkable generative capability, but their sequential nature results in significant inference latency. Existing training-free acceleration methods typically verify tokens independently, overlooking the strong co-occurrence patterns between adjacent visual tokens. This independence assumption often leads to contextual inconsistency and limits decoding efficiency. In this work, we introduce a novel training-free acceleration framework that performs phrase-level speculative verification, enabling the model to jointly validate multiple correlated tokens within each decoding window. To construct such phrase units, we analyze token co-occurrence statistics from the training corpus and group frequently co-occurring tokens into semantically coherent visual phrases. During inference, the proposed phrase-level verification evaluates aggregated likelihood ratios over each phrase, allowing simultaneous acceptance of multiple tokens while preserving generation quality. Extensive experiments on autoregressive text-to-image generation show that our method significantly reduces the number of function evaluations (NFE) and achieves up to 30% faster decoding without compromising visual fidelity. Our findings reveal that modeling short-range token co-occurrence provides an effective and general principle for accelerating autoregressive inference.