search

pathak22 / context-encoder

[CVPR 2016] Unsupervised Feature Learning by Image Inpainting using GANs
https://www.cs.cmu.edu/~dpathak/context_encoder/
Other
876 stars 204 forks source link

Why there is no batch normalization after the first conv2d layer? #27

Open ghost opened 6 years ago

pathak22 commented 6 years ago

@include-chenshen No particular reason as such -- just wanted to keep Discriminator, Generator, and Encoder as symmetric as possible. I do not think this choice should affect the results in any way. Feel free to try it out. :-)

ghost commented 6 years ago

Oh, I am so sorry! Because of some reason, I missed this e-mail until today. Since you shared your precious dataset with me, I have done train process. But, limited by computational power, I am unable to train on the whole imagenet. So, could you share the checkpointing of discriminator (corresponding to “paris_inpaintCenter.t7” or “imagenet_inpaintCenter.t7” on Github) with me? Anyway, thank you again.

On 17 Feb 2018, 3:34 PM +0800, Deepak Pathak notifications@github.com, wrote:

@include-chenshen No particular reason as such -- just wanted to keep Discriminator, Generator, and Encoder as symmetric as possible. I do not think this choice should affect the results in any way. Feel free to try it out. :-) — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.