LANTERN: Accelerating Visual Autoregressive Models with Relaxed Speculative Decoding

TL;DR

LANTERN introduces a relaxed decoding method that significantly accelerates visual autoregressive models by addressing token selection ambiguity, enabling more flexible token use without sacrificing image quality.

Contribution

The paper proposes LANTERN, a novel relaxed decoding approach that improves speculative decoding efficiency in visual AR models by leveraging token interchangeability in latent space.

Findings

LANTERN achieves 1.75x speed-up over naive speculative decoding.

LANTERN achieves 1.82x speed-up over greedy decoding.

The method maintains image quality and semantic coherence.

Abstract

Auto-Regressive (AR) models have recently gained prominence in image generation, often matching or even surpassing the performance of diffusion models. However, one major limitation of AR models is their sequential nature, which processes tokens one at a time, slowing down generation compared to models like GANs or diffusion-based methods that operate more efficiently. While speculative decoding has proven effective for accelerating LLMs by generating multiple tokens in a single forward, its application in visual AR models remains largely unexplored. In this work, we identify a challenge in this setting, which we term \textit{token selection ambiguity}, wherein visual AR models frequently assign uniformly low probabilities to tokens, hampering the performance of speculative decoding. To overcome this challenge, we propose a relaxed acceptance condition referred to as LANTERN that leverages the interchangeability of tokens in latent space. This relaxation restores the effectiveness of speculative decoding in visual AR models by enabling more flexible use of candidate tokens that would otherwise be prematurely rejected. Furthermore, by incorporating a total variation distance bound, we ensure that these speed gains are achieved without significantly compromising image quality or semantic coherence. Experimental results demonstrate the efficacy of our method in providing a substantial speed-up over speculative decoding. In specific, compared to a na\"ive application of the state-of-the-art speculative decoding, LANTERN increases speed-ups by $\mathbf{1.75}\times$ and $\mathbf{1.82}\times$, as compared to greedy decoding and random sampling, respectively, when applied to LlamaGen, a contemporary visual AR model. The code is publicly available at https://github.com/jadohu/LANTERN.