deepfakes_faceswap it's OOM if i use tensorflow-gpu. Did i do something wrong?

it's OOM if i use tensorflow-gpu. Did i do something wrong?

Open JiayunLin opened this issue 7 years ago • 2 comments

I use 1060 with 3G vram, and it's useless....... train_on_batch() use a lot of memory Did i do something wrong?

Dec 27 '17 08:12 JiayunLin

Same with GTX 960 4gb....

Jan 25 '18 21:01 Rnijland

Fixed it, the model.py has a formula that allocates 8gb ram. Probably someone with a 1080gtx made it.

Cant paste the right code here, as star symbols won`t show... Follow This:

https://pastebin.com/7dyWECmT

Jan 25 '18 22:01 Rnijland