How to test the trained generator models?

[kalai2033]

I have completed the training successfully using Image2Image.py. But I could not test the model using test images. Whenever i run with --eval , the model generates the inference for random images in the training dataset. Could you please help me with it?

Also in the eval() function,

generate_images function is hard coded with the value of 100. Why is that so?

[f90]

So you are saying everything works well except it generates predictions for the train dataset? I don't know why that would be the case, I set it up to use the validation dataset here, a few lines before the generate_images call you cited:

# DATASET
dataset = get_aligned_dataset(opt, "val")
input_dataset = CropDataset(dataset, lambda x: x[0:dataset.A_nc, :, :])

I hard-coded the number of images arbitrarily to 100 to have a decent number of outputs to look at, but you can freely change it without breaking the code in any way.