heykeetae/Self-Attention-GAN
Pytorch implementation of Self-Attention Generative Adversarial Networks (SAGAN)
Python
Issues
- 3
the meaning of Gamma in Attention model
#57 opened by hank19960918 - 0
the download.sh can't use
#67 opened by hanjiahe1999 - 0
- 0
- 4
Cannot Download CelebA dataset
#39 opened by StellaLI-0501 - 0
Add examples to work with audio files as well
#64 opened by tipycalFlow - 0
Trying the Self-Attention-GAN with dog images
#63 opened by liku-amare - 3
torch.bmm(), CUDA out of memory.
#59 opened by lyricgoal - 8
- 4
- 1
- 0
- 1
Does this code supports multi-class datasets?
#18 opened by babyjie57 - 0
Negative self.gamma parameter??
#60 opened by theelon - 2
The code is different from the original paper
#54 opened by lmisssunl - 11
About the attention map
#19 opened by o0t1ng0o - 1
permute twice?
#37 opened by aflyingnoob - 1
- 2
is there no activation on self attn layer?
#40 opened by caffeinism - 4
new data set
#30 opened by asifkyni - 6
I add self_Attn to another network , but it causes cuda error : out of memory
#29 opened by cooodeKnight - 0
- 10
About negative gamma
#47 opened by SeokjuLee - 0
RuntimeError: cublas runtime error : the GPU program failed to execute at /pytorch/aten/src/THC/THCBlas.cu:450
#55 opened by RokiaAbdeen - 2
FID and IS
#34 opened by Avyukth - 4
- 1
dropbox link missing
#52 opened by NisargBhatt73 - 1
model.py is working only for imsize=64
#49 opened by franknb - 4
- 1
dataset structure
#46 opened by Tianshuo-Xu - 0
detach fake image when updating the discriminator
#48 opened by kam1107 - 2
- 3
test class missing
#35 opened by enochkan - 0
about visualization code
#44 opened by amiltonwong - 0
Weird Error while using multi GPU.
#43 opened by AIprogrammer - 2
- 0
Softmax
#41 opened by abelmouhcine - 0
- 0
Variable image size.
#36 opened by caffeinism - 0
class conditional batch norm not implemented
#33 opened by djl11 - 0
Gumbel Softmax?
#32 opened by emited - 0
Spectral norm on Generator
#31 opened by dougsouza - 1
- 0
I want to train the model by cifar-10 using batch size=128, but there is negative gamma emerged. Is it influence the results generated?
#28 opened by long-username - 0
Hello,heykeetae,has this implement implied conditional batch normalization? It seems that not support multi-class image generation.
#26 opened by mrj-taffy - 1
hinge loss
#20 opened by IPNUISTlegal - 0
How to use in 3D conv?
#24 opened by guanfaqian - 0
Which Loss for CelebA Results?
#22 opened by tegillis - 0
- 0
d_iters seems not being used
#21 opened by r06922019