List your company
ResearchFormerly WizWand

Likelihood-based Generative Modeling on OMNIGLOT

86.04NLL (nats)Generalized GBN
84.328895.8794107.43118.9806Aug 6, 2024
Updated 4mo ago

Evaluation results

MethodLinks
Generalized GBNModel type=Gamma VAE Models, Importance samples=10002024.0886.04
VDVAEModel type=VAE Models with an Unconditional Decoder, Importance samples=10002024.0886.93
Attention VAEModel type=VAE Models with an Unconditional Decoder, Importance samples=10002024.0889.5
Lossy VAEModel type=VAE and Flow Models with Autoregressive Components, Importance samples=10002024.0889.83
NVAEModel type=VAE Models with an Unconditional Decoder, Importance samples=10002024.0890.18
Conv DRAWModel type=VAE Models with an Unconditional Decoder, Importance samples=10002024.0891
BIVAModel type=VAE Models with an Unconditional Decoder, Importance samples=10002024.0893.54
DirVAEModel type=Non-Gaussian VAE Models, Importance samples=10002024.0895.82
DVAE++Model type=VAE Models with an Unconditional Decoder, Importance samples=10002024.0897.43
SBVAEModel type=Non-Gaussian VAE Models, Importance samples=10002024.08128.82