Number of images produce by GAN

@Sebgolos Do we have control over number of images produce by GAN?
Let say I have dataset of 60 images and i want GAN to produce 30 times more images than original. How could i approach this?

Just set a batch of latent vectors to be 30*60?
Albeit this would probably make your GPU explode.

I don’t see a problem in generating 180 times a batch of 10 fake images from latent vectors (sampled again each time).

fixed_latent = torch.randn(10, latent_size, 1, 1, device=device)
taking Epochs = 180.
I think this is what you are trying to say??

Not sure if I would call it epochs.

You first train the GAN (not sure how long - as long as necessary).
Then you have a generator, which you can ask for a generated images.

Just ask it many times and it should give you any amount of images you want.