List your company
ResearchFormerly WizWand

Generative Modeling on CIFAR-10

2.51BPDCR-NVAE
2.46722.75613.0453.3339Jul 9, 2018Jul 14, 2019Jul 18, 2020Jul 23, 2021Jul 28, 2022Aug 2, 2023Aug 6, 2024
Updated 4mo ago

Evaluation results

MethodLinks
CR-NVAEResolution=32 x 322021.052.51
NVAE+AugResolution=32 x 322021.052.7
Attention VAEModel type=VAE Models with an Unconditional Decoder, Importance samples=1002024.082.79
δ-VAEResolution=32x322020.072.83
δ-VAEModel type=VAE and Flow Models with Autoregressive Components2024.082.83
Generalized GBNModel type=Gamma VAE Models, Variational layers (L)=332024.082.84
PixelSNAILModel family=Autoregressive2019.022.85
PixelSNAILResolution=32x322020.072.85
PixelSNAILModel type=Autoregressive Models2024.082.85
VDVAEModel type=VAE Models with an Unconditional Decoder, Variational layers (L)=332024.082.87
Image TransformerModel family=Autoregressive2019.022.9
PixelVAE++Resolution=32x322020.072.9
Image TransformerResolution=32x322020.072.9
PixelVAE++Model type=VAE and Flow Models with Autoregressive Components2024.082.9
Image TransformerModel type=Autoregressive Models2024.082.9
NVAEResolution=32x32, Flow=With flow2020.072.91
NVAEResolution=32 x 322021.052.91
NVAEModel type=VAE Models with an Unconditional Decoder2024.082.91
PixelCNN++Model family=Autoregressive2019.022.92
PixelCNN++Resolution=32x322020.072.92
PixelCNN++Model type=Autoregressive Models2024.082.92
NVAEResolution=32x32, Flow=Without flow2020.072.93
MAEResolution=32x322020.072.95
Lossy VAEResolution=32x322020.072.95
MAEModel type=VAE and Flow Models with Autoregressive Components2024.082.95
Lossy VAEModel type=VAE and Flow Models with Autoregressive Components2024.082.95
VFlowResolution=32x322020.072.98
VFlowModel type=Flow Models without any Autoregressive Components2024.082.98
PixelRNNModel family=Autoregressive2019.023
PixelRNNResolution=32x322020.073
PixelRNNModel type=Autoregressive Models2024.083
Gated PixelCNNModel family=Autoregressive2019.023.03
Gated PixelCNNResolution=32x322020.073.03
Gated PixelCNNModel type=Autoregressive Models2024.083.03
ANFResolution=32x322020.073.05
ANFModel type=Flow Models without any Autoregressive Components2024.083.05
Flow++Model family=Non-autoregressive2019.023.08
BIVAResolution=32x322020.073.08
Flow++Resolution=32x322020.073.08
BIVAModel type=VAE Models with an Unconditional Decoder2024.083.08
Flow++Model type=Flow Models without any Autoregressive Components2024.083.08
IAF-VAEModel family=Non-autoregressive2019.023.11
IAF-VAEResolution=32x322020.073.11
PixelCNNModel family=Autoregressive2019.023.14
MaCowResolution=32x322020.073.16
MaCowModel type=VAE and Flow Models with Autoregressive Components2024.083.16
MALIFlow Type=Continuous Flow (FFJORD)2021.023.27
Residual flowResolution=32x322020.073.28
Flow++Flow Type=Discrete Flow2021.023.28
Residual FlowFlow Type=Discrete Flow2021.023.28
Residual flowModel type=Flow Models without any Autoregressive Components2024.083.28
GlowK=32, L=3, permutation=invertible 1 x 1 convolution2018.073.35
GlowModel family=Non-autoregressive2019.023.35
GLOWResolution=32x322020.073.35
SemiNormFlow Type=Continuous Flow (FFJORD)2021.023.35
GlowFlow Type=Discrete Flow2021.023.35
DVAE++Resolution=32x322020.073.38
RNODEFlow Type=Continuous Flow (FFJORD)2021.023.38
DVAE++Model type=VAE Models with an Unconditional Decoder2024.083.38
VanillaFlow Type=Continuous Flow (FFJORD)2021.023.4
i-ResNetFlow Type=Discrete Flow2021.023.45
RealNVPK=32, L=32018.073.49
RealNVPModel family=Non-autoregressive2019.023.49
Real NVPResolution=32x322020.073.49
RealNVPFlow Type=Discrete Flow2021.023.49
Real NVPModel type=Flow Models without any Autoregressive Components2024.083.49
Conv DrawResolution=32x322020.073.58
Conv DRAWModel type=VAE Models with an Unconditional Decoder2024.083.58