NVlabs / edm

Elucidating the Design Space of Diffusion-Based Generative Models (EDM)

NVlabs/edm Issues

Using EDM for end-to-end training of latent diffusion models
Updated 3 days ago
Question about Fourier Embedding and EDM Loss calculation
Closed 22 days ago1
Question about parameter tuning
Updated a month ago1
Is it possible to also share the code for Figure 1b (ideal denoiser outputs) in the paper
Updated 3 months ago
ERROR when using multi-gpu training
Closed 8 months ago7
AssertionError: Torch not compiled with CUDA enabled
Closed 5 months ago
About hyper-parameters selection
Updated 5 months ago
Different Phases of Sampling
Updated 6 months ago
Can not reproduct the result in Table 2.
Updated 6 months ago5
On checkpoint for 256*256 images
Updated 6 months ago
Question about Augmentation
Updated 8 months ago1
How to select options shown in Table2 for training VP/VE models
Updated 9 months ago1
VP checkpoints are trained using VE-scaling?
Closed a year ago2
params[x] in this process with sizes [x, x] appears not to match sizes of the same param in process 0
Closed a year ago2
Could you please provide us imagenet-uncond checkpoint?
Updated 10 months ago
About the Checkpoint
Updated a year ago
Is it possible to also share the code for Figure 3 in the paper
Closed a year ago1
Is the original loss of VP/VE implemented (i.e. network predict epsilon)?
Closed a year ago1
DataLoader worker (pid xxxx) is killed by signal
Closed a year ago1
.
Closed a year ago
Wrong beta_D coefficient for VP during sampling?
Updated 2 years ago1
Deterministic sampling and sampling steps
Updated 2 years ago
Zero initialization of convolutions
Updated 2 years ago1
Possible error in the up-sampling function
Closed 2 years ago1
Question about the parameterizations of VP-SDE
Closed 2 years ago1