search

zsyOAOA / DifFace

DifFace: Blind Face Restoration with Diffused Error Contraction (TPAMI, 2024)
Other
628 stars 42 forks source link

training time #3

Closed 123456789-qwer closed 1 year ago

123456789-qwer commented 1 year ago

How long it takes to train the entire network?

zsyOAOA commented 1 year ago

To train the diffusion model, 8 V100 GPU, taking about one week.

To train the diffused estimator based on SwinIR, 4 V100 GPUs, taking about 4 days.

@123456789-qwer