search

dansuh17 / segan-pytorch

SEGAN pytorch implementation https://arxiv.org/abs/1703.09452
GNU General Public License v3.0
106 stars 32 forks source link

The output of discriminator is [1 x 8]? #3

Closed simpleoier closed 6 years ago

simpleoier commented 6 years ago

As a true or false indicator, shouldn't the output of discriminator be a scalar?

dansuh17 commented 6 years ago

@simpleoier yes, I guess you're right. I think I was mistaken by

in the last activation layer thereis a one-dimensional convolution layer with one filter of width one that does not downsample the hidden activations (1 × 1 convolution).

in the paper. But it seems clear from the code that there is a fully connected layer required before the output of discriminator.

Thanks for commenting; I'll put up a commit soon that fixes this :) (or you can open a pull request, if you don't mind)

simpleoier commented 6 years ago

Actually, in the paper

The latter (3) reduces the amount of parameters re- quired for the final classification neuron, which is fully con- nected to all hidden activations with a linear behavior.

dansuh17 commented 6 years ago

I'll keep this open until the fix has been made

dansuh17 commented 6 years ago

Fix made with commit 26f2a8e. closing.