Thanks for your sharing. I have one question: how about using DDIM in your proposed method? In the "ddim_sample" function, I notice when I use the "sample=out["mean"]", the final output is meaningful. However, when I use "sample=mean_pred", the final output is just a noise map.
Thanks for your sharing. I have one question: how about using DDIM in your proposed method? In the "ddim_sample" function, I notice when I use the "sample=out["mean"]", the final output is meaningful. However, when I use "sample=mean_pred", the final output is just a noise map.