odegeasslbc / self-supervised-sketch-to-image-synthesis-pytorch Goto Github PK

hi, i make an error : [Errno 2] No such file or directory: 'pretrained/models/12000.pth'

where should i find it?

PRETRAINED_AE_ITER = 12000 in config.py

Hi!
Can you release the pre training model of Tom? During the training, I found that my Tom model could not achieve the good results in the paper.
thanks a lot!

Hello, your creative work is very interesting!
Have you released pre-trained models? I expect to get your pretrained model. thanks a lot.
Looking forward to your recovery!

Hello, There are some matters when I try to train the Sketch-to-image model with your code.
At line 22 of the file "train_step_2_gan.py", I couldn't find the file "generate_matrix".I found the function with the same name in the other two files in the folder "evaluate". I don't know which is the right.

I can't understand why the program reports an error after calculating the PID.

I need your help very much, thank you！

Hi,

Thanks for your awesome work, i am trying to reproduce your result currently but encounter some issues.
I keep getting cuda out of memory error when execute train_step_1_ae.py.

My setting is

gpu: 1080ti (11.441MiB memory)
batch size: 2
image size: 512

Moreover, i change the DepthwiseConv2d in ResBlkG to nn.Conv2d because i can not correclty install torch_dwconv, i am not sure whether it has great influence or not.

I saw that you mention in the other issue that we can easily reproduce the result by single GPU. I am not sure where i might be wrong. Can you help me ?

Are pretrained models available? Can you upload them to Google drive？Thanks!

Hello! I'm having some issues with generating images using the model. You said that there is a code for sketch-to-image in folder "evaluate", but I didn't find it, which code should I use to generate the image? Thanks!