denoising-diffusion-pytorch

denoising-diffusion-pytorch copied to clipboard

I'm getting the following error since the update from 2 days ago when loading a previously trained model: I can keep working by using the previous version: "!pip install denoising_diffusion_pytorch==0.31"...

hizzlehoff

I am doing this locally but it would be great if the package had this small change to allow sampling at different resolutions: Line 627 in denoising_diffusion_pytorch.py: @torch.no_grad() def sample(self,...

hizzlehoff

Sharing answer from the question "How many iterations do I need?"

4

**[Training Code]** ``` model = Unet( dim = 64, dim_mults = (1, 2, 4, 8) ).cuda() diffusion = GaussianDiffusion( model, image_size = 128, timesteps = 1000, # number of steps...

SeungyounShin

Hi, thanks for your amazing code. I am wondering whether the diffusion model can handle input with different sizes during training. For example, if I have some training images with...

dannalily

A question on time embedding

1

 I don't understand why (half_dim - 1) is used here. According to the transformer's time-coding formula, there should be "emb = math.log(10000) / half_dim", I don't think half_dim should...

shixishi

How to derive the equation of alpha_cosine_log_snr

1

Thanks for your excellent project! I am still confused about how to derive the equation of alpha_cosine_log_snr after reading IDDPM and VDM, can somebody help? https://github.com/lucidrains/denoising-diffusion-pytorch/blob/dfbafee555bdae80b55d63a989073836bbfc257e/denoising_diffusion_pytorch/continuous_time_gaussian_diffusion.py#L69

winnechan

Question about version of pytorch

2

Can I ask what is the version of pytorch? I tried to use pytorch 1.9.0-cu102 it didn't work. Plus, why this code is citing DDIM? Is this a modified DDPM?...

ghost

training on own dataset but start getting black images after "sample-10"

2

Thank you sharing the wonderful code! I was using my own dataset for training but start getting black images after 9000 steps. Anyone knows how to fix this situation? Here...

twentyfiveYang

No available kernel

1

I'm running into the "No available kernel" problem during training, has anyone else run into the same problem as me and how did they resolve it? Thanks for the answer!

xianghuahua

Hi, is it possible to use optimized attention in windows somehow? Pytorch prints a warning that flash attention is missing, it is not working in windows pytorch. Can one use...

reinterpret-cast