denoising-diffusion-pytorch

denoising-diffusion-pytorch copied to clipboard

Hi, First of all thank you so much for providing excellent codes. Do you have any plan to provide the evaluation file?

vivek231

Technical question about sampling function

2

Hi, I was wondering why every diffusion models implementation uses this specific sampling procedure? When I take a look at the DDPM paper they show the sampling algorithm to be:...

dome272

"pred_x0" preforms better than "pred_noise"

8

Hi, guys. I have conducted experiments on MNIST and FashionMNIST. In my results, training with "objective == pred_x0" achieves better results than "objective == pred_noise" on both datasets, which is...

TheSunWillRise

tuttyfrutyee

Why are ema updates not reflected on the online model?

3

As far as I understood from the code, the ema updates are affecting only the final data sampler rather than the online model that is being trained. If the assumption...

tuttyfrutyee

Hi, thanks your code , the paper is said that diffusion model can not reverse the image so , how to reconstruction input image like Fig8 in paper ??

Johnson-yue

How long does it takes to train.

24

Thanks for your clean implementation sharing. I try on celeba datasets. After 150k steps, the generated images are not well as it claimed in the paper and the flowers you...

qsh-zh

GPU utilization

5

With the unoptimized implementation on github.com/openai/guided-diffusion, a GPU utilization of about 20 to 30% can be archived during training. What percentatge can be archived with this implementation?

AndreasBergmeister

Any suggestion to generate 1-D vectors using diffusion model

2

Recently I am trying to use the diffusion model to do 1-D vector generation task, such as to generate sentence embedding which is originally generated from Bert, I have some...

animebing

How do epochs work with ddpm?

1

For example, if I set 700k training steps, and stop at ~30k when there are 15k images, is there an idea of how many times the training data has been...

sgbaird