jychoi118 / ilvr_adm

ILVR: Conditioning Method for Denoising Diffusion Probabilistic Models (ICCV 2021 Oral)
MIT License
417 stars 51 forks source link

Questions about the training process #27

Closed zy-charon closed 1 year ago

zy-charon commented 1 year ago

Hi, thank you very much for your great work. I have trained a model with class_cond as true before, and using this model to directly sample has a torch size mismatch problem. I would like to ask if there are any special requirements for parameter setting if training a diffusion model from scratch. Is it enough to keep the parameter class_cond as false and other parameters to be reasonable? error

jychoi118 commented 1 year ago

Try setting --learn_sigma True when training the model. The output channel of the model is 6 instead of 3, because the model learns both epsilon and sigma.

zy-charon commented 1 year ago

Try setting --learn_sigma True when training the model. The output channel of the model is 6 instead of 3, because the model learns both epsilon and sigma.

Thank you, I'll try again