Speculative Jacobi-Denoising Decoding for Accelerating Autoregressive Text-to-image Generation
This addresses inefficiency in text-to-image generation for users needing faster visual content creation, though it is incremental as it builds on existing autoregressive models.
The paper tackles the slow inference of autoregressive text-to-image models by proposing Speculative Jacobi-Denoising Decoding (SJD2), which enables parallel token generation through denoising and Jacobi iterations, reducing model forward passes while maintaining image quality.
As a new paradigm of visual content generation, autoregressive text-to-image models suffer from slow inference due to their sequential token-by-token decoding process, often requiring thousands of model forward passes to generate a single image. To address this inefficiency, we propose Speculative Jacobi-Denoising Decoding (SJD2), a framework that incorporates the denoising process into Jacobi iterations to enable parallel token generation in autoregressive models. Our method introduces a next-clean-token prediction paradigm that enables the pre-trained autoregressive models to accept noise-perturbed token embeddings and predict the next clean tokens through low-cost fine-tuning. This denoising paradigm guides the model towards more stable Jacobi trajectories. During inference, our method initializes token sequences with Gaussian noise and performs iterative next-clean-token-prediction in the embedding space. We employ a probabilistic criterion to verify and accept multiple tokens in parallel, and refine the unaccepted tokens for the next iteration with the denoising trajectory. Experiments show that our method can accelerate generation by reducing model forward passes while maintaining the visual quality of generated images.