Parallel Rollout Approximation for Pixel-Space Autoregressive Image Generation

• URL: http://arxiv.org/abs/2606.27978v1
• Date: Fri, 26 Jun 2026 11:27:39 GMT
• Title: Parallel Rollout Approximation for Pixel-Space Autoregressive Image Generation
• Abstract summary: We propose emphParallel Rollout Approximation (PRA), a scalable framework that addresses both challenges jointly.<n>PRA generates low-dimensional intermediate states instead of high-dimensional pixel patches, then maps them back to pixel-space tokens with a pixel decoder, preserving a pixel-in, pixel-out AR interface.<n>On class-conditional ImageNet-1K generation at $256times256$ resolution, PRA-S with 135M parameters achieves an FID of 2.58, surpassing the previous billion-scale pixel-space AR result of 3.60.
• Score: 9.30873702701651
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Pixel-space continuous-token autoregressive (AR) generation directly models images as sequences of raw pixel patches, avoiding discrete tokenization or a separately pretrained tokenizer. However, it faces coupled challenges: high-dimensional patch generation causes large single-step errors, and teacher-forced training creates a train--inference gap that makes these errors accumulate across AR steps. Existing fixes such as $x$-prediction and input noise injection only partially mitigate these issues. Exact rollout training better matches inference-time conditions, but is impractical due to prohibitively slow sequential sampling. We propose \emph{Parallel Rollout Approximation} (PRA), a scalable framework that addresses both challenges jointly. PRA generates low-dimensional intermediate states instead of high-dimensional pixel patches, then maps them back to pixel-space tokens with a pixel decoder, preserving a pixel-in, pixel-out AR interface. It also constructs inference-like pixel inputs through the same intermediate-state-to-pixel path used at inference, independently across positions, approximating the pixel-feedback interface encountered during inference-time rollout while retaining parallel teacher-forced training. On class-conditional ImageNet-1K generation at $256\times256$ resolution, PRA-S with 135M parameters achieves an FID of 2.58, surpassing the previous billion-scale pixel-space AR result of 3.60. Scaling to PRA-L with 511M parameters further improves FID to 1.94, establishing a new state of the art among pixel-space AR models. Beyond generation, PRA achieves higher ImageNet classification probing accuracy than other AR and diffusion baselines, suggesting its potential for unified pixel-space image generation and understanding.

Related papers

• DPAR: Dynamic Patchification for Efficient Autoregressive Visual Generation [10.719563134726057]
  We present DPAR, a decoder-only autoregressive model that aggregates image tokens into a variable number of patches for efficient image generation.<n>DPAR reduces token count by 1.81x and 2.06x on Imagenet 256 and 384 generation resolution respectively, leading to a reduction of up to 40% FLOPs in training costs.
  arXiv  Detail & Related papers  (2025-12-26T05:03:47Z)
• VA-$π$: Variational Policy Alignment for Pixel-Aware Autoregressive Generation [65.22452273252428]
  VA-$$ is a post-training framework to optimize autoregressive visual generation.<n>It unifies pixel reconstruction and autoregressive modeling.<n>It reduces FID from 14.36 to 7.65 and improves IS from 86.55 to 116.70 on LlamaGen-XXL.
  arXiv  Detail & Related papers  (2025-12-22T18:54:30Z)
• Representation Entanglement for Generation: Training Diffusion Transformers Is Much Easier Than You Think [63.25744258438214]
  REPA and its variants effectively mitigate training challenges in diffusion models by incorporating external visual representations from pretrained models.<n>We argue that the external alignment, which is absent during the entire denoising inference process, falls short of fully harnessing the potential of discriminative representations.<n>We propose Representation Entanglement for Generation (REG), which entangles low-level image latents with a single high-level class token from pretrained foundation models for denoising.
  arXiv  Detail & Related papers  (2025-07-02T08:29:18Z)
• Token-Shuffle: Towards High-Resolution Image Generation with Autoregressive Models [92.18057318458528]
  Token-Shuffle is a novel method that reduces the number of image tokens in Transformer.<n>Our strategy requires no additional pretrained text-encoder and enables MLLMs to support extremely high-resolution image synthesis.<n>In GenAI-benchmark, our 2.7B model achieves 0.77 overall score on hard prompts, outperforming AR models LlamaGen by 0.18 and diffusion models LDM by 0.15.
  arXiv  Detail & Related papers  (2025-04-24T17:59:56Z)
• Accelerating Image Super-Resolution Networks with Pixel-Level Classification [29.010136088811137]
  Pixel-level for Single Image SuperResolution is a novel method designed to distribute computational resources adaptively at the pixel level. Our method allows for performance and computational cost balance during inference without re-training.
  arXiv  Detail & Related papers  (2024-07-31T08:53:10Z)
• CoordFill: Efficient High-Resolution Image Inpainting via Parameterized Coordinate Querying [52.91778151771145]
  In this paper, we try to break the limitations for the first time thanks to the recent development of continuous implicit representation. Experiments show that the proposed method achieves real-time performance on the 2048$times$2048 images using a single GTX 2080 Ti GPU.
  arXiv  Detail & Related papers  (2023-03-15T11:13:51Z)
• PixelFolder: An Efficient Progressive Pixel Synthesis Network for Image Generation [88.55256389703082]
  Pixel is a promising research paradigm for image generation, which can well exploit pixel-wise prior knowledge for generation. In this paper, we propose a progressive pixel synthesis network towards efficient image generation, as Pixel. With much less expenditure, Pixel obtains new state-of-the-art (SOTA) performance on two benchmark datasets.
  arXiv  Detail & Related papers  (2022-04-02T10:55:11Z)
• Parallel Discrete Convolutions on Adaptive Particle Representations of Images [2.362412515574206]
  We present data structures and algorithms for native implementations of discrete convolution operators over Adaptive Particle Representations. The APR is a content-adaptive image representation that locally adapts the sampling resolution to the image signal. We show that APR convolution naturally leads to scale-adaptive algorithms that efficiently parallelize on multi-core CPU and GPU architectures.
  arXiv  Detail & Related papers  (2021-12-07T09:40:05Z)
• LAPAR: Linearly-Assembled Pixel-Adaptive Regression Network for Single Image Super-Resolution and Beyond [75.37541439447314]
  Single image super-resolution (SISR) deals with a fundamental problem of upsampling a low-resolution (LR) image to its high-resolution (HR) version. This paper proposes a linearly-assembled pixel-adaptive regression network (LAPAR) to strike a sweet spot of deep model complexity and resulting SISR quality.
  arXiv  Detail & Related papers  (2021-05-21T15:47:18Z)
• Adversarial Generation of Continuous Images [31.92891885615843]
  In this paper, we propose two novel architectural techniques for building INR-based image decoders. We use them to build a state-of-the-art continuous image GAN. Our proposed INR-GAN architecture improves the performance of continuous image generators by several times.
  arXiv  Detail & Related papers  (2020-11-24T11:06:40Z)
• Locally Masked Convolution for Autoregressive Models [107.4635841204146]
  LMConv is a simple modification to the standard 2D convolution that allows arbitrary masks to be applied to the weights at each location in the image. We learn an ensemble of distribution estimators that share parameters but differ in generation order, achieving improved performance on whole-image density estimation.
  arXiv  Detail & Related papers  (2020-06-22T17:59:07Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.