After reading replies to one of the issues, I used the guided diffusion repository, to train 2 DDPMs. All the steps except 1 are clear to me. Regarding the yaml file(like afhq.yaml), are the parameters in it supposed to be equal to the ones used to train the original models( like beta_start, beta_end).
Along with this, should i use i_DDPM function or DDPM class ?
Thanks
After reading replies to one of the issues, I used the guided diffusion repository, to train 2 DDPMs. All the steps except 1 are clear to me. Regarding the yaml file(like afhq.yaml), are the parameters in it supposed to be equal to the ones used to train the original models( like beta_start, beta_end).
Along with this, should i use i_DDPM function or DDPM class ?
Thanks