Malek
Malek
Is there any update on this? In my experience this detail has been crucial in determining sample quality, yet it seems to be largely unaddressed with regards to diffusion models....
Hey @TheSunWillRise, I have also been having trouble training diffusion models using the pred_noise objective, and I wasn't sure if it was something on my end or some issue with...
> Hey @TheSunWillRise, I have also been having trouble training diffusion models using the pred_noise objective, and I wasn't sure if it was something on my end or some issue...
@guozhiyao , unfortunately not. You have to implement clamping when using pred_noise objective, though no clamping is necessary with pred_x0 objective. Overall, I still found that with clamping, the pred_noise...
Hey @dome272 , I am not sure why the code framework here does not work with the equation you referenced in the paper, and I have not had time to...
@Smith42 Would you be able to show some samples/results from training your CelebA model? It seems that a lot of other people are struggling to reproduce the results shown in...
Has anyone been able to apply this code for tasks like inpainting or super-resolution? They are both forms of conditional generation, and it's not clear to me how one should...
> > @xenova But that should be a final step, not done on every iteration. > > The problem I'm experiencing is that the reverse process doesn't converge toward the...
Does anyone have any updates on why this is not working?