Image Tokenizer Needs Post-Training

• URL: http://arxiv.org/abs/2509.12474v1
• Date: Mon, 15 Sep 2025 21:38:03 GMT
• Title: Image Tokenizer Needs Post-Training
• Authors: Kai Qiu, Xiang Li, Hao Chen, Jason Kuen, Xiaohao Xu, Jiuxiang Gu, Yinyi Luo, Bhiksha Raj, Zhe Lin, Marios Savvides,
• Abstract summary: We propose a novel tokenizer training scheme, focusing on improving latent space construction and decoding respectively.<n>Specifically, we propose a plug-and-play tokenizer training scheme, which significantly enhances the robustness of tokenizer.<n>We further optimize the tokenizer decoder regarding a well-trained generative model to mitigate the distribution difference between generated and reconstructed tokens.
• Score: 76.91832192778732
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Recent image generative models typically capture the image distribution in a pre-constructed latent space, relying on a frozen image tokenizer. However, there exists a significant discrepancy between the reconstruction and generation distribution, where current tokenizers only prioritize the reconstruction task that happens before generative training without considering the generation errors during sampling. In this paper, we comprehensively analyze the reason for this discrepancy in a discrete latent space, and, from which, we propose a novel tokenizer training scheme including both main-training and post-training, focusing on improving latent space construction and decoding respectively. During the main training, a latent perturbation strategy is proposed to simulate sampling noises, \ie, the unexpected tokens generated in generative inference. Specifically, we propose a plug-and-play tokenizer training scheme, which significantly enhances the robustness of tokenizer, thus boosting the generation quality and convergence speed, and a novel tokenizer evaluation metric, \ie, pFID, which successfully correlates the tokenizer performance to generation quality. During post-training, we further optimize the tokenizer decoder regarding a well-trained generative model to mitigate the distribution difference between generated and reconstructed tokens. With a $\sim$400M generator, a discrete tokenizer trained with our proposed main training achieves a notable 1.60 gFID and further obtains 1.36 gFID with the additional post-training. Further experiments are conducted to broadly validate the effectiveness of our post-training strategy on off-the-shelf discrete and continuous tokenizers, coupled with autoregressive and diffusion-based generators.

Related papers

• SFTok: Bridging the Performance Gap in Discrete Tokenizers [72.9996757048065]
  We propose textbfSFTok, a discrete tokenizer that incorporates a multi-step iterative mechanism for precise reconstruction.<n>At a high compression rate of only 64 tokens per image, SFTok achieves state-of-the-art reconstruction quality on ImageNet.
  arXiv  Detail & Related papers  (2025-12-18T18:59:04Z)
• Repulsor: Accelerating Generative Modeling with a Contrastive Memory Bank [65.00301565190824]
  mname is a plug-and-play training framework that requires no external encoders.<n>mname achieves a state-of-the-art FID of textbf2.40 within 400k steps, significantly outperforming comparable methods.
  arXiv  Detail & Related papers  (2025-12-09T14:39:26Z)
• Supervised Contrastive Learning for Few-Shot AI-Generated Image Detection and Attribution [3.103291412074661]
  This work proposes a novel two-stage detection framework to address the generalization challenge inherent in synthetic image detection.<n>The proposed framework achieves an average detection accuracy of 91.3%, representing a 5.2 percentage point improvement over existing approaches.
  arXiv  Detail & Related papers  (2025-11-20T16:53:24Z)
• MixAR: Mixture Autoregressive Image Generation [12.846100277592969]
  We introduce MixAR, a novel framework that injects discrete tokens as prior guidance for continuous autoregressive modeling.<n>We investigate several discrete-continuous mixture strategies, including self-attention (DC-SA), cross-attention (DC-CA), and a simple approach (DC-Mix) that replaces homogeneous mask tokens with informative discrete counterparts.
  arXiv  Detail & Related papers  (2025-11-15T12:19:28Z)
• BIGFix: Bidirectional Image Generation with Token Fixing [21.40682276355247]
  We propose a method for self-correcting image generation by iteratively refining sampled tokens.<n>We achieve this with a novel training scheme that injects random tokens in the context, improving robustness and enabling token fixing during sampling.<n>We evaluate our approach on image generation using the ImageNet-256 and CIFAR-10 datasets, as well as on video generation with UCF-101 and NuScenes, demonstrating substantial improvements across both modalities.
  arXiv  Detail & Related papers  (2025-10-14T07:34:44Z)
• Speculative Jacobi-Denoising Decoding for Accelerating Autoregressive Text-to-image Generation [110.28291466364784]
  Speculative Jacobi-Denoising Decoding (SJD2) is a framework that incorporates the denoising process into Jacobi to enable parallel token generation in autoregressive models.<n>Our method introduces a next-clean-token prediction paradigm that enables the pre-trained autoregressive models to accept noise-perturbed token embeddings.
  arXiv  Detail & Related papers  (2025-10-10T04:30:45Z)
• Joint Learning of Pose Regression and Denoising Diffusion with Score Scaling Sampling for Category-level 6D Pose Estimation [19.996312940214363]
  We propose a novel pipeline for category-level 6D object pose estimation.<n>The proposed method pretrains the encoder with the direct pose regression head, and jointly learns the networks via the regression head and the denoising diffusion head.<n>The sampling guidance maintains multi-modal characteristics of symmetric objects at early denoising steps while ensuring high-quality pose generation at final steps.
  arXiv  Detail & Related papers  (2025-10-05T09:58:51Z)
• Robust Latent Matters: Boosting Image Generation with Sampling Error Synthesis [57.7367843129838]
  Recent image generation schemes typically capture image distribution in a pre-constructed latent space relying on a frozen image tokenizer.<n>We propose a novel plug-and-play tokenizer training scheme to facilitate latent space construction.
  arXiv  Detail & Related papers  (2025-03-11T12:09:11Z)
• Continuous Speculative Decoding for Autoregressive Image Generation [33.05392461723613]
  Continuous-valued Autoregressive (AR) image generation models have demonstrated notable superiority over their discrete-token counterparts. speculative decoding has proven effective in accelerating Large Language Models (LLMs) This work generalizes the speculative decoding algorithm from discrete tokens to continuous space.
  arXiv  Detail & Related papers  (2024-11-18T09:19:15Z)
• Faster Language Models with Better Multi-Token Prediction Using Tensor Decomposition [5.575078692353885]
  We propose a new model for multi-token prediction in transformers, aiming to enhance sampling efficiency without compromising accuracy.<n>By generalizing it to a rank-$r$ canonical probability decomposition, we develop an improved model that predicts multiple tokens simultaneously.
  arXiv  Detail & Related papers  (2024-10-23T11:06:36Z)
• Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusion [61.03681839276652]
  Diffusion Forcing is a new training paradigm where a diffusion model is trained to denoise a set of tokens with independent per-token noise levels.<n>We apply Diffusion Forcing to sequence generative modeling by training a causal next-token prediction model to generate one or several future tokens.
  arXiv  Detail & Related papers  (2024-07-01T15:43:25Z)
• Latent Autoregressive Source Separation [5.871054749661012]
  This paper introduces vector-quantized Latent Autoregressive Source Separation (i.e., de-mixing an input signal into its constituent sources) without requiring additional gradient-based optimization or modifications of existing models. Our separation method relies on the Bayesian formulation in which the autoregressive models are the priors, and a discrete (non-parametric) likelihood function is constructed by performing frequency counts over latent sums of addend tokens.
  arXiv  Detail & Related papers  (2023-01-09T17:32:00Z)

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.