Can we train the DG and backbone jointly?

Question

Can we train the DG and backbone jointly?

WelkinYang opened this issue a year ago · 2 comments

Thanks for your great contribution! And training DG individually requires pre-generation of a large number of samples using diffusion models which is time-consuming. Can we train the DG and a converged diffusion model jointly?

Answer 1 · 2023-04-25T06:40:31.000Z

Please see the camera-ready version on simultaneous training of discriminator and score networks. It will be disclosed within a month.

Answer 2 · 2023-04-27T05:11:59.000Z

We have just uploaded the camera-ready version of the paper. Please see the Discussion section at https://arxiv.org/abs/2211.17091 for additional information on simultaneous training.