Research map / Image generation / Autoregressive and latent generation
Generating Diverse High-Fidelity Images with VQ-VAE-2
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Evaluate the classifier on VQ-VAE reconstructions instead of generated samples | VQ-VAE · ImageNet | Top-1 Accuracy: 54.83 -> 58.74 | Helped |
| Evaluate the classifier on VQ-VAE reconstructions instead of generated samples | VQ-VAE · ImageNet | Top-5 Accuracy: 77.59 -> 80.98 | Helped |
| Apply classifier-based rejection sampling | VQ-VAE · ImageNet | FID: roughly ~30 to ~10 | Helped |
| Apply classifier-based rejection sampling | VQ-VAE · ImageNet | Inception Score | Helped |
| Use VQ-VAE samples rather than BigGAN-deep samples | ImageNet · ImageNet | Recall | Helped |
| Use VQ-VAE samples rather than BigGAN-deep samples | ImageNet · ImageNet | Precision | Hurt |
Closest papers by meaning
- Diffusion bridges vector quantized variational autoencoders (ICML 2022 · similarity 0.91)
- From Variational to Deterministic Autoencoders (ICLR 2020 · similarity 0.90)
- Adversarial Symmetric Variational Autoencoder (NeurIPS 2017 · similarity 0.89)
- MoVQ: Modulating Quantized Vectors for High-Fidelity Image Generation (NeurIPS 2022 · similarity 0.89)
- EQ-VAE: Equivariance Regularized Latent Space for Improved Generative Image Modeling (ICML 2025 · similarity 0.89)
- Rethinking the Objectives of Vector-Quantized Tokenizers for Image Synthesis (CVPR 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.