mistake in Gumbel softmax sample function.

YongfeiYan / Gumbel_Softmax_VAE

PyTorch implementation of a Variational Autoencoder with Gumbel-Softmax Distribution

198 stars 37 forks source link

mistake in Gumbel softmax sample function. #3

Closed sobalgi closed 5 years ago

sobalgi commented 5 years ago

The gumbel_softmax_sample function is logits + gumbel_sample. But it should have been F.log_softmax(logits) + gumbel_sample according to the paper. Is this not a mistake?