sudo docker pull taesungp/pytorch-cyclegan-and-pix2pix
sudo nvidia-docker run -it -p 8097:8097 taesungp/pytorch-cyclegan-and-pix2pix bash
cd /workspace/pytorch-CycleGAN-and-pix2pix
bash datasets/download_pix2pix_dataset.sh facades
cd /workspace/scripts
vim train_pix2pix.sh
增加:--gpu_ids 0,1,2 --batch_size 2 -niter 100000
python -m visdom.server &bash scripts/train_pix2pix.sh
一般情况下 visdom会报错,或者长时间没有反应
cd /miniconda/lib/python2.7/site-packages/visdom/
vim server.py
注掉:download_scripts()
#download_scripts()
之后运行 ,看GPU状态
这里运行可能会出现 ,port used的错误。
lsof -i:8097
kill PID(pid下面的那个号)
之后再python -m visdom.server &bash scripts/train_pix2pix.sh
应该就没问题