This may be the simplest implement of DDPM. You can directly run Main.py to train the UNet on CIFAR-10 dataset and see the amazing process of denoising.
Hi
for evaluating with FID score in the conditional model, should we compare e.g. 1k or 10k images from each class between original and the generated ones?
e.g. FID( real(car), generated(car) ) FID( real(elephant), generated(elephant) ) etc ?
Hi for evaluating with FID score in the conditional model, should we compare e.g. 1k or 10k images from each class between original and the generated ones? e.g. FID( real(car), generated(car) ) FID( real(elephant), generated(elephant) ) etc ?