VibeToken: Scaling 1D Image Tokenizers and Autoregressive Models for Dynamic Resolution Generations
This work addresses the efficiency and resolution flexibility bottlenecks in autoregressive image generation, narrowing the gap with diffusion models for production use cases.
VibeToken introduces a resolution-agnostic 1D image tokenizer and autoregressive generator that achieves 3.94 gFID on 1024x1024 images using only 64 tokens, outperforming a diffusion baseline (5.87 gFID, 1024 tokens) while being 63.4x more compute-efficient than fixed-resolution AR models like LlamaGen.
We introduce an efficient, resolution-agnostic autoregressive (AR) image synthesis approach that generalizes to arbitrary resolutions and aspect ratios, narrowing the gap to diffusion models at scale. At its core is VibeToken, a novel resolution-agnostic 1D Transformer-based image tokenizer that encodes images into a dynamic, user-controllable sequence of 32-256 tokens, achieving a state-of-the-art efficiency and performance trade-off. Building on VibeToken, we present VibeToken-Gen, a class-conditioned AR generator with out-of-the-box support for arbitrary resolutions while requiring significantly fewer compute resources. Notably, VibeToken-Gen synthesizes 1024x1024 images using only 64 tokens and achieves 3.94 gFID; by comparison, a diffusion-based state-of-the-art alternative requires 1,024 tokens and attains 5.87 gFID. In contrast to fixed-resolution AR models such as LlamaGen -- whose inference FLOPs grow quadratically with resolution (11T FLOPs at 1024x1024) -- VibeToken-Gen maintains a constant 179G FLOPs (63.4x efficient) independent of resolution. We hope VibeToken can help unlock the wide adoption of AR visual generative models in production use cases.