Salvato in:
Dettagli Bibliografici
Autori principali: Xu, Tongda, Zheng, Wendi, He, Jiajun, Hernandez-Lobato, Jose Miguel, Wang, Yan, Zhang, Ya-Qin, Tang, Jie
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2512.06609
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911717300109312
author Xu, Tongda
Zheng, Wendi
He, Jiajun
Hernandez-Lobato, Jose Miguel
Wang, Yan
Zhang, Ya-Qin
Tang, Jie
author_facet Xu, Tongda
Zheng, Wendi
He, Jiajun
Hernandez-Lobato, Jose Miguel
Wang, Yan
Zhang, Ya-Qin
Tang, Jie
contents Vector-quantized variational autoencoders (VQ-VAEs) are discrete autoencoders that compress images into discrete tokens. However, they are difficult to train due to discretization. In this paper, we propose a simple yet effective technique dubbed Gaussian Quant (GQ), which first trains a Gaussian VAE under certain constraints and then converts it into a VQ-VAE without additional training. For conversion, GQ generates random Gaussian noise as a codebook and finds the closest noise vector to the posterior mean. Theoretically, we prove that when the logarithm of the codebook size exceeds the bits-back coding rate of the Gaussian VAE, a small quantization error is guaranteed. Practically, we propose a heuristic to train Gaussian VAEs for effective conversion, named the target divergence constraint (TDC). Empirically, we show that GQ outperforms previous VQ-VAEs, such as VQGAN, FSQ, LFQ, and BSQ, on both UNet and ViT architectures. Furthermore, TDC also improves previous Gaussian VAE discretization methods, such as TokenBridge. The source code is provided in https://github.com/tongdaxu/VQ-VAE-from-Gaussian-VAE.
format Preprint
id arxiv_https___arxiv_org_abs_2512_06609
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Training-Free Vector Quantization via Gaussian VAEs
Xu, Tongda
Zheng, Wendi
He, Jiajun
Hernandez-Lobato, Jose Miguel
Wang, Yan
Zhang, Ya-Qin
Tang, Jie
Machine Learning
Computer Vision and Pattern Recognition
Vector-quantized variational autoencoders (VQ-VAEs) are discrete autoencoders that compress images into discrete tokens. However, they are difficult to train due to discretization. In this paper, we propose a simple yet effective technique dubbed Gaussian Quant (GQ), which first trains a Gaussian VAE under certain constraints and then converts it into a VQ-VAE without additional training. For conversion, GQ generates random Gaussian noise as a codebook and finds the closest noise vector to the posterior mean. Theoretically, we prove that when the logarithm of the codebook size exceeds the bits-back coding rate of the Gaussian VAE, a small quantization error is guaranteed. Practically, we propose a heuristic to train Gaussian VAEs for effective conversion, named the target divergence constraint (TDC). Empirically, we show that GQ outperforms previous VQ-VAEs, such as VQGAN, FSQ, LFQ, and BSQ, on both UNet and ViT architectures. Furthermore, TDC also improves previous Gaussian VAE discretization methods, such as TokenBridge. The source code is provided in https://github.com/tongdaxu/VQ-VAE-from-Gaussian-VAE.
title Training-Free Vector Quantization via Gaussian VAEs
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.06609