Clarification for ddim_sample_loop_known

[bennymi]

Hi, thank you for providing the code. I have a question about the ddim_sample_loop_known function. Why do we set the timesteps in this line (

Diffusion-based-Segmentation/guided_diffusion/gaussian_diffusion.py

Line 792 in 676f214

t = th.randint(499,500, (b,), device=device).long().to(device)

) to be 499?

Shouldn't the SpacedDiffusion model and the timestep_respacing with something like ddim 50 automatically take care of the correct beta scheduling to reduce the number of timesteps?

In the original ddim_sample_loop implementation, there is no such thing: https://github.com/openai/improved-diffusion/blob/1bc7bbbdc414d83d4abf2ad8cc1446dc36c4e4d5/improved_diffusion/gaussian_diffusion.py#L562