But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...A vector quantization library originally transcribed from Deepmind's tensorflow implementation, made conveniently into a package. It uses exponential moving averages to update the dictionary. VQ has been successfully used by Deepmind and OpenAI for high quality generation of images (VQ-VAE-2) and music (Jukebox). We first propose multiple improvements over vanilla VQGAN from architecture to codebook learning, yielding better efficiency and reconstruction fidelity. The improved ViT-VQGAN further improves vector-quantized image modeling tasks, including unconditional, class-conditioned image generation and unsupervised representation learning.We present SoundStream, a novel neural audio codec that can efficiently compress speech, music and general audio at bitrates normally targeted by speech-tailored codecs. SoundStream relies on a model architecture composed by a fully convolutional encoder/decoder network and a residual vector quantizer, which are trained jointly end-to-end. Training leverages recent advances in text-to-speech ...Vector-quantized Image Modeling with Improved VQGAN Jiahui Yu, Xin Li, Jing Yu Koh, Han Zhang, Ruoming Pang, James Qin, Alexander Ku, Yuanzhong Xu, Jason Baldridge, Yonghui Wu ICLR 2022. BEiT v2: Masked Image Modeling with Vector-Quantized Visual Tokenizers Zhiliang Peng, Li Dong, Hangbo Bao, Qixiang Ye, Furu Wei arXiv 2022.In “ Vector-Quantized Image Modeling with Improved VQGAN ”, we propose a two-stage model that reconceives traditional image quantization techniques to yield improved performance on image generation and image understanding tasks. In the first stage, an image quantization model, called VQGAN, encodes an image into lower-dimensional discrete ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...Vector-Quantized Image Modeling with ViT-VQGAN. One recent, commonly used model that quantizes images into integer tokens is the Vector-quantized Variational AutoEncoder (VQVAE), a CNN-based auto-encoder whose latent space is a matrix of discrete learnable variables, trained end-to-end.The release weight of ViT-VQGAN small which is trained on ImageNet at here; 16/08. First release weight of ViT-VQGAN base which is trained on ImageNet at here; Add an colab notebook at here; About The Project. This is an unofficial implementation of both ViT-VQGAN and RQ-VAE in Pytorch. ViT-VQGAN is a simple ViT-based Vector Quantized ...Vector-quantized Image Modeling with Improved VQGAN Jiahui Yu, Xin Li, Jing Yu Koh, Han Zhang, Ruoming Pang, James Qin, Alexander Ku, Yuanzhong Xu, Jason Baldridge, Yonghui Wu ICLR 2022. BEiT v2: Masked Image Modeling with Vector-Quantized Visual Tokenizers Zhiliang Peng, Li Dong, Hangbo Bao, Qixiang Ye, Furu Wei arXiv 2022.Vector-Quantized Image Modeling with ViT-VQGAN One recent, commonly used model that quantizes images into integer tokens is the Vector-quantized Variational AutoEncoder (VQVAE), a CNN-based auto-encoder whose latent space is a matrix of discrete learnable variables, trained end-to-end. VQGAN is an improved version of this that introduces an ...Our experiments show that causal decoder-only models trained on an autoregressive language modeling objective exhibit the strongest zero-shot generalization after purely unsupervised pretraining. However, models with non-causal visibility on their input trained with a masked language modeling objective followed by multitask finetuning perform ...论文标题：《Vector-Quantized Image Modeling with Improved VQGAN》—— ICLR 2022 作者信息：Jiahui Yu等 Google Research 这篇论文提出了VQGAN这样的模型不仅可以应用在图像生成中，其预训练模型还可以通过微调迁移到图像分类等任务中去。Vector-quantized Image Modeling with Improved VQGAN Jiahui Yu, Xin Li, Jing Yu Koh, Han Zhang, Ruoming Pang, James Qin, Alexander Ku, Yuanzhong Xu, Jason Baldridge, Yonghui Wu ICLR 2022. BEiT v2: Masked Image Modeling with Vector-Quantized Visual Tokenizers Zhiliang Peng, Li Dong, Hangbo Bao, Qixiang Ye, Furu Wei arXiv 2022.But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...and Yonghui Wu. Vector-quantized image modeling with improved vqgan. arXiv preprint arXiv:2110.04627, 2021.3 [10]Chuanxia Zheng, Long Tung Vuong, Jianfei Cai, and Dinh Phung. Movq: Modulating quantized vectors for high-fidelity image generation.arXiv preprint arXiv:2209.09002, 2022.3But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...Oct 9, 2021 · Motivated by this success, we explore a Vector-quantized Image Modeling (VIM) approach that involves pretraining a Transformer to predict rasterized image tokens autoregressively. The... Described as “a bunch of Python that can take words and make pictures based on trained data sets," VQGANs (Vector Quantized Generative Adversarial Networks) pit neural networks against one another to synthesize “plausible” images. Much coverage has been on the unsettling applications of GANs, but they also have benign uses. Hands-on access through a simplified front-end helps us develop ...Motivated by this success, we explore a Vector-quantized Image Modeling (VIM) approach that involves pretraining a Transformer to predict rasterized image tokens autoregressively. The discrete image tokens are encoded from a learned Vision-Transformer-based VQGAN (ViT-VQGAN). We first propose multiple improvements over vanilla VQGAN from ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...Motivated by this success, we explore a Vector-quantized Image Modeling (VIM) approach that involves pretraining a Transformer to predict rasterized image tokens autoregressively. The...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization techniques to yield improved performance on image generation and image understanding tasks.In “ Vector-Quantized Image Modeling with Improved VQGAN ”, we propose a two-stage model that reconceives traditional image quantization techniques to yield improved performance on image generation and image understanding tasks. In the first stage, an image quantization model, called VQGAN, encodes an image into lower-dimensional discrete ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...Image-Text Pre-training with Contrastive Captioners ... 2022 Vector-Quantized Image Modeling with Improved VQGAN مئی 17, 2022 Contextual Rephrasing in Google ...DALL-E 2 - Pytorch. Implementation of DALL-E 2, OpenAI's updated text-to-image synthesis neural network, in Pytorch.. Yannic Kilcher summary | AssemblyAI explainer. The main novelty seems to be an extra layer of indirection with the prior network (whether it is an autoregressive transformer or a diffusion network), which predicts an image embedding based on the text embedding from CLIP.The release weight of ViT-VQGAN small which is trained on ImageNet at here; 16/08. First release weight of ViT-VQGAN base which is trained on ImageNet at here; Add an colab notebook at here; About The Project. This is an unofficial implementation of both ViT-VQGAN and RQ-VAE in Pytorch. ViT-VQGAN is a simple ViT-based Vector Quantized ...1 code implementation • 29 May 2023 • Zi Wang , Alexander Ku , Jason Baldridge , Thomas L. Griffiths , Been Kim. Our experiments show it can (1) probe a model's representations of concepts even with a very small number of examples, (2) accurately measure both epistemic uncertainty (how confident the probe is) and aleatory uncertainty (how ...Vector-quantized image modeling with improved vqgan J Yu, X Li, JY Koh, H Zhang, R Pang, J Qin, A Ku, Y Xu, J Baldridge, Y Wu The Tenth International Conference on Learning Representations , 2021A vector quantization library originally transcribed from Deepmind's tensorflow implementation, made conveniently into a package. It uses exponential moving averages to update the dictionary. VQ has been successfully used by Deepmind and OpenAI for high quality generation of images (VQ-VAE-2) and music (Jukebox). Vector-Quantized Image Modeling with ViT-VQGAN. One recent, commonly used model that quantizes images into integer tokens is the Vector-quantized Variational AutoEncoder (VQVAE), a CNN-based auto-encoder whose latent space is a matrix of discrete learnable variables, trained end-to-end.Vision transformers (ViTs) have gained popularity recently. Even without customized image operators such as convolutions, ViTs can yield competitive performance when properly trained on massive data. However, the computational overhead of ViTs remains prohibitive, due to stacking multi-head self-attention modules and else. Compared to the vast literature and prevailing success in compressing ...We first propose multiple improvements over vanilla VQGAN from architecture to codebook learning, yielding better efficiency and reconstruction fidelity. The improved ViT-VQGAN further improves vector-quantized image modeling tasks, including unconditional, class-conditioned image generation and unsupervised representation learning.Vector-quantized Image Modeling with Improved VQGAN Jiahui Yu, Xin Li, Jing Yu Koh, Han Zhang, Ruoming Pang, James Qin, Alex Ku, Yuanzhong Xu, Jason Baldridge, Yonghui Wu ICLR 2022 / Google AI Blog. SimVLM: Simple Visual Language Model Pretraining with Weak Supervision Zirui Wang, Jiahui Yu, Adams Wei Yu, Zihang Dai, Yulia Tsvetkov, Yuan CaoDescribed as “a bunch of Python that can take words and make pictures based on trained data sets," VQGANs (Vector Quantized Generative Adversarial Networks) pit neural networks against one another to synthesize “plausible” images. Much coverage has been on the unsettling applications of GANs, but they also have benign uses. Hands-on access through a simplified front-end helps us develop ...Described as “a bunch of Python that can take words and make pictures based on trained data sets," VQGANs (Vector Quantized Generative Adversarial Networks) pit neural networks against one another to synthesize “plausible” images. Much coverage has been on the unsettling applications of GANs, but they also have benign uses. Hands-on access through a simplified front-end helps us develop ...Semantic image synthesis enables control over unconditional image generation by allowing guidance on what is being generated. We conditionally synthesize the latent space from a vector quantized model (VQ-model) pre-trained to autoencode images. Instead of training an autoregressive Transformer on separately learned conditioning latents and ...This strategy can naturally tap into the rich body of prior work on large language models, which have seen continued advances in capabilities and performance through scaling data and model sizes. Our approach is simple: First, Parti uses a Transformer-based image tokenizer, ViT-VQGAN, to encode images as sequences of discrete tokens.Prior works have largely connected image to text through pretraining and/or fine-tuning on curated image-text datasets, which can be a costly and expensive process. In order to resolve this limitation, we propose a simple yet effective approach called Language-Quantized AutoEncoder (LQAE), a modification of VQ-VAE that learns to align text ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...Posted by Jiahui Yu, Senior Research Scientist, and Jing Yu Koh, Research Software Engineer, Google Research In recent years, natural language processing models have dramatically improved their ability to learn general-purpose representations, which has resulted in significant performance gains for a wide range of natural language generation and natural language understanding tasks. In large ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...Described as “a bunch of Python that can take words and make pictures based on trained data sets," VQGANs (Vector Quantized Generative Adversarial Networks) pit neural networks against one another to synthesize “plausible” images. Much coverage has been on the unsettling applications of GANs, but they also have benign uses. Hands-on access through a simplified front-end helps us develop ...Posted by Jiahui Yu, Senior Research Scientist, and Jing Yu Koh, Research Software Engineer, Google Research In recent years, natural language processing models have dramatically improved their ability to learn general-purpose representations, which has resulted in significant performance gains for a wide range of natural language generation and natural language understanding tasks. In large ...一、改进点： 1.stage1（image quantization ViT-VQGAN）： 基于ViT的VQGAN encoder。 基于VQGAN做了从架构到码本学习方式的多种改进——>提升了efficiency和reconstruction fidelity. 包括logits-laplace loss，L2 loss，adversarial loss 和 perceptual loss. 2.stage2（vector-quantized image modeling VIM）： 学习了一个自回归的transformer，包括无条件生成/类条件生成/无监督表示学习。Posted by Jiahui Yu, Senior Research Scientist, and Jing Yu Koh, Research Software Engineer, Google Research In recent years, natural language processing models have dramatically improved their ability to learn general-purpose representations, which has resulted in significant performance gains for a wide range of natural language generation and natural language understanding tasks. In large ...The improved ViT-VQGAN further improves vector-quantized image modeling tasks, including unconditional, class-conditioned image generation and unsupervised representation learning. When trained on ImageNet at 256x256 resolution, we achieve Inception Score (IS) of 175.1 and Fr'echet Inception Distance (FID) of 4.17, a dramatic improvement over ...In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization techniques to yield improved performance on image generation and image understanding tasks.Prior works have largely connected image to text through pretraining and/or fine-tuning on curated image-text datasets, which can be a costly and expensive process. In order to resolve this limitation, we propose a simple yet effective approach called Language-Quantized AutoEncoder (LQAE), a modification of VQ-VAE that learns to align text ...Semantic image synthesis enables control over unconditional image generation by allowing guidance on what is being generated. We conditionally synthesize the latent space from a vector quantized model (VQ-model) pre-trained to autoencode images. Instead of training an autoregressive Transformer on separately learned conditioning latents and ...Abstract and Figures. Although two-stage Vector Quantized (VQ) generative models allow for synthesizing high-fidelity and high-resolution images, their quantization operator encodes similar ...Vector-Quantized Image Modeling with ViT-VQGAN One recent, commonly used model that quantizes images into integer tokens is the Vector-quantized Variational AutoEncoder (VQVAE), a CNN-based auto-encoder whose latent space is a matrix of discrete learnable variables, trained end-to-end.But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...Vector-Quantized Image Modeling with Improved VQGAN maj 17, 2022 ... Image-Text Pre-training with Contrastive Captioners ... Vector-Quantized Image Modeling with ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...arXiv.org e-Print archiveVector-quantized Image Modeling with Improved VQGAN Yu, Jiahui ; Li, Xin ; Koh, Jing Yu ; Zhang, Han ; Pang, Ruoming ; Qin, James ; Ku, Alexander ; Xu, Yuanzhong Prior works have largely connected image to text through pretraining and/or fine-tuning on curated image-text datasets, which can be a costly and expensive process. In order to resolve this limitation, we propose a simple yet effective approach called Language-Quantized AutoEncoder (LQAE), a modification of VQ-VAE that learns to align text ...Motivated by this success, we explore a Vector-quantized Image Modeling (VIM) approach that involves pretraining a Transformer to predict rasterized image tokens autoregressively. The discrete image tokens are encoded from a learned Vision-Transformer-based VQGAN (ViT-VQGAN). We first propose multiple improvements over vanilla VQGAN from ...openreview.net あくまで個人的なメモVQGANの改善とベクトル量子化を使った画像生成モデル・画像分類モデルの改善。VQVAEはCNNベースのAE、VQGANはそこにadversarial lossを導入した。 これらはCNNのauto encoder(AE)の学習(ステージ1)とencodeしたlatent variablesの密度をCNN(or Transformer)で学習する(ステージ2)という2つ ...一、改进点： 1.stage1（image quantization ViT-VQGAN）： 基于ViT的VQGAN encoder。 基于VQGAN做了从架构到码本学习方式的多种改进——>提升了efficiency和reconstruction fidelity. 包括logits-laplace loss，L2 loss，adversarial loss 和 perceptual loss. 2.stage2（vector-quantized image modeling VIM）： 学习了一个自回归的transformer，包括无条件生成/类条件生成/无监督表示学习。Motivated by this success, we explore a Vector-quantized Image Modeling (VIM) approach that involves pretraining a Transformer to predict rasterized image tokens autoregressively. The...Autoregressive Image Generation using Residual Quantization ...Vector-Quantized Image Modeling with ViT-VQGAN. One recent, commonly used model that quantizes images into integer tokens is the Vector-quantized Variational AutoEncoder (VQVAE), a CNN-based auto-encoder whose latent space is a matrix of discrete learnable variables, trained end-to-end.Our experiments show that causal decoder-only models trained on an autoregressive language modeling objective exhibit the strongest zero-shot generalization after purely unsupervised pretraining. However, models with non-causal visibility on their input trained with a masked language modeling objective followed by multitask finetuning perform ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...Vector-Quantized Image Modeling with ViT-VQGAN. One recent, commonly used model that quantizes images into integer tokens is the Vector-quantized Variational AutoEncoder (VQVAE), a CNN-based auto-encoder whose latent space is a matrix of discrete learnable variables, trained end-to-end.Abstract and Figures. Although two-stage Vector Quantized (VQ) generative models allow for synthesizing high-fidelity and high-resolution images, their quantization operator encodes similar ...Vector-Quantized Image Modeling with Improved VQGAN may 17, 2022 ... Image-Text Pre-training with Contrastive Captioners ... Vector-Quantized Image Modeling with ...Motivated by this success, we explore a Vector-quantized Image Modeling (VIM) approach that involves pretraining a Transformer to predict rasterized image tokens autoregressively. The discrete image tokens are encoded from a learned Vision-Transformer-based VQGAN (ViT-VQGAN).Abstract and Figures. Although two-stage Vector Quantized (VQ) generative models allow for synthesizing high-fidelity and high-resolution images, their quantization operator encodes similar ...Vector-Quantized Image Modeling with ViT-VQGAN. One recent, commonly used model that quantizes images into integer tokens is the Vector-quantized Variational AutoEncoder (VQVAE), a CNN-based auto-encoder whose latent space is a matrix of discrete learnable variables, trained end-to-end.
Vector-quantized Image Modeling with Improved VQGAN Yu, Jiahui ; Li, Xin ; Koh, Jing Yu ; Zhang, Han ; Pang, Ruoming ; Qin, James ; Ku, Alexander ; Xu, Yuanzhong. Rain bird transformer lowepercent27s
The discrete image tokens are encoded from a learned Vision-Transformer-based VQGAN (ViT-VQGAN). We first propose multiple improvements over vanilla VQGAN from architecture to codebook learning, yielding better efficiency and reconstruction fidelity. The improved ViT-VQGAN further improves vector-quantized image modeling tasks, including ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...Vector-Quantized Image Modeling with ViT-VQGAN. One recent, commonly used model that quantizes images into integer tokens is the Vector-quantized Variational AutoEncoder (VQVAE), a CNN-based auto-encoder whose latent space is a matrix of discrete learnable variables, trained end-to-end.Our experiments show that causal decoder-only models trained on an autoregressive language modeling objective exhibit the strongest zero-shot generalization after purely unsupervised pretraining. However, models with non-causal visibility on their input trained with a masked language modeling objective followed by multitask finetuning perform ...We describe multiple improvements to the image quantizer and show that training a stronger image quantizer is a key component for improving both image generation and image understanding. Vector-Quantized Image Modeling with ViT-VQGAN One recent, commonly used model that quantizes images into integer tokens is the Vector-quantized Variational ...Motivated by this success, we explore a Vector-quantized Image Modeling (VIM) approach that involves pretraining a Transformer to predict rasterized image tokens autoregressively. The discrete image tokens are encoded from a learned Vision-Transformer-based VQGAN (ViT-VQGAN).But while such models have achieved strong performance for image generation, few studies have evaluated the learned representation for downstream discriminative tasks (such as image classification). In “Vector-Quantized Image Modeling with Improved VQGAN”, we propose a two-stage model that reconceives traditional image quantization ...Image encoders compress an image into smaller dimensions, sometimes even quantized into a discrete space (such as the VQGAN from taming-transformers used in Craiyon). In this article, we try to reproduce the results from ViT-VQGAN (" Vector-quantized Image Modeling with Improved VQGAN ") and experiment with further adaptations.“Vector-Quantized Image Modeling with Improved VQGAN” proposes a two-stage model that reinvents classic image quantization methods to produce better picture generation and image understanding tasks. The first step is to encode an image into discrete latent codes of lesser dimensions using an image quantization model called VQGAN.Vector-Quantized Image Modeling with Improved VQGAN maj 17, 2022 ... Image-Text Pre-training with Contrastive Captioners ... Vector-Quantized Image Modeling with ...论文标题：《Vector-Quantized Image Modeling with Improved VQGAN》—— ICLR 2022 作者信息：Jiahui Yu等 Google Research 这篇论文提出了VQGAN这样的模型不仅可以应用在图像生成中，其预训练模型还可以通过微调迁移到图像分类等任务中去。Vector-Quantized Image Modeling with ViT-VQGAN. One recent, commonly used model that quantizes images into integer tokens is the Vector-quantized Variational AutoEncoder (VQVAE), a CNN-based auto-encoder whose latent space is a matrix of discrete learnable variables, trained end-to-end..