HyperAIHyperAI

Image Generation On Imagenet 32X32

Metrics

bpd

Results

Performance results of various models on this benchmark

Model Name
bpd
Paper TitleRepository
NVAE w/ flow3.92NVAE: A Deep Hierarchical Variational Autoencoder-
Glow (Kingma and Dhariwal, 2018)4.09Glow: Generative Flow with Invertible 1x1 Convolutions-
MintNet4.06MintNet: Building Invertible Neural Networks with Masked Convolutions-
Residual Flow4.01Residual Flows for Invertible Generative Modeling-
VDM3.72Variational Diffusion Models-
SPN Menick and Kalchbrenner (2019)3.85Generating High Fidelity Images with Subscale Pixel Networks and Multidimensional Upscaling-
StyleGAN-XL-StyleGAN-XL: Scaling StyleGAN to Large Diverse Datasets-
δ-VAE3.77Preventing Posterior Collapse with delta-VAEs-
PaGoDA-PaGoDA: Progressive Growing of a One-Step Generator from a Low-Resolution Diffusion Teacher-
Very Deep VAE3.8Very Deep VAEs Generalize Autoregressive Models and Can Outperform Them on Images-
PixelRNN3.86Pixel Recurrent Neural Networks-
Hourglass3.74Hierarchical Transformers Are More Efficient Language Models-
DDPM++ (VP, NLL) + ST3.85Soft Truncation: A Universal Training Technique of Score-based Diffusion Model for High Precision Score Estimation-
i-DODE3.43Improved Techniques for Maximum Likelihood Estimation for Diffusion ODEs-
MRCNF3.77Multi-Resolution Continuous Normalizing Flows-
Flow++3.86Flow++: Improving Flow-Based Generative Models with Variational Dequantization and Architecture Design-
BIVA Maaloe et al. (2019)3.96BIVA: A Very Deep Hierarchy of Latent Variables for Generative Modeling-
Reflected Diffusion3.74Reflected Diffusion Models-
NDM3.55Neural Diffusion Models-
DDPM3.89Denoising Diffusion Probabilistic Models-
0 of 33 row(s) selected.
Image Generation On Imagenet 32X32 | SOTA | HyperAI