Dataset used for training ffhq_10m.pt

jychoi118 / ilvr_adm

ILVR: Conditioning Method for Denoising Diffusion Probabilistic Models (ICCV 2021 Oral)

MIT License

417 stars 51 forks source link

Dataset used for training ffhq_10m.pt #6

Closed sunyasheng closed 2 years ago

sunyasheng commented 2 years ago

Thanks for your amazing work and nice sharing. You've said that 10M images are used to train ffhq_10m.pt in your REAME. As far as I know, there are 70000 images in FFHQ dataset. It is a little confusing. So is there any other data I've missed?

jychoi118 commented 2 years ago

Yes, there are 70000 images in FFHQ. What I meant for 10M is a number of images seen by the model, which is batch_size x iteration, or epoch x 70000.

sunyasheng commented 2 years ago

Thank you very much for quick reply. That makes sense. Since it clarifies my question, I will close this issue.