search

conormdurkan / neural-statistician

PyTorch Implementation of Neural Statistician
58 stars 12 forks source link

Inference Network Question #1

Closed na-dong closed 5 years ago

na-dong commented 5 years ago

Hello, thanks for sharing code. While reading Inference Network part, I want to know about the reasone that you add (z,c,h) instead of concatenating as mentioned in the paper pseudo code. Thanks.

conormdurkan commented 5 years ago

There’s no particular reason, and I don’t think there’s much practical difference between the two. One advantage of addition over concatenation is that it saves on parameters. You could try out concatenation, but I’m pretty sure it won't change anything.

On 5 Oct 2018, at 08:56, na-dong <notifications@github.com mailto:notifications@github.com> wrote:

Hello, thanks for sharing code. While reading Inference Network part, I want to know about the reasone that you add (z,c,h) instead of concatenating as mentioned in the paper pseudo code. Thanks.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/conormdurkan/neural-statistician/issues/1, or mute the thread https://github.com/notifications/unsubscribe-auth/AKPEcg5lDWchJldCJXgdKkty_6cilebXks5uhxC7gaJpZM4XJvE5.

na-dong commented 5 years ago

Thanks for the reply.