search

zsyOAOA / DifFace

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

adjust_lr() missing 1 required argument #6

Open mshaheryar91 opened 1 year ago

mshaheryar91 commented 1 year ago

adjust_lr() missing 1 required positional argument: 'ii'adjust_lr() missing 1 required positional argument: 'ii'.

In trainer_py trainer.py", line 255, in train self.adjust_lr()self.adjust_lr()

Please correct me if I am wrong? What to put there? I am trying to train model with batch size [8, 2]

zsyOAOA commented 1 year ago

I have corrected this problem. Please have a look again. Thanks.

mshaheryar91 commented 1 year ago

Can you specify which line have you changed? I will download that file only

mshaheryar91 commented 1 year ago

Ok Dear i got the file, its train.py and i update the code and it's running well. Thanks. Can you please specify on 4, 12GB GPU's roughly how much time it will take to train?

HWalkingMan commented 1 year ago

I can only run 1k iterations per 12min on the 3090, so I need at least 160h to train the full 800k iterations and 23.9G of GPU memory. Can you please publish your training time?

chensming commented 1 year ago

I can only run 1k iterations per 12min on the 3090, so I need at least 160h to train the full 800k iterations and 23.9G of GPU memory. Can you please publish your training time?

hi, bro, how many gpus do you use to train? (1 or 8?)

zsyOAOA commented 1 year ago

I train the diffusion model with 8 V100 GPUs, about taking one week. Note that you can only implement 500k iteration, having similar performance. @chensming @HWalkingMan