Skip to content

Latest commit

 

History

History
75 lines (51 loc) · 2.34 KB

installation.md

File metadata and controls

75 lines (51 loc) · 2.34 KB

Installation

Python 3.6

Must use python 3.6 because tensorflow-gpu is not available for python 3.7 yet.

Python virtualenv

requirements.txt containt the necessary requirements and the CPU versions of TensorFlow and pyTorch. See below for more details about instaling both with GPU support.

pip install virtualenv
virtualenv -p python3 .env
cd .env/Scripts
activate
cd ..
cd ..
pip install -r requirements.txt

Using GPU

Requires an NVIDIA GPU with compute capability >= 3.0. Link.

Install TensorFlow

Follow instructions on GPU Support page and installation page which gets down to this:

  • install NVidia Driver
  • install CUDA 9.0 + PATH configurations
  • install cuDNN
  • pip install tensorflow-gpu

Install pyTorch

pip install http://download.pytorch.org/whl/cu90/torch-0.4.1-cp36-cp36m-win_amd64.whl
pip install torchvision

Using CPU

Install TensorFlow

pip install tensorflow

Install pyTorch

pip install http://download.pytorch.org/whl/cpu/torch-0.4.1-cp36-cp36m-win_amd64.whl
pip install torchvision

Running StarGAN

CPU measures

The following command will need 11GB of RAM and 140 days to complete! Notice that batch_size and log_step affects the needed memory.

python main.py --mode train --dataset CelebA --image_size 128 --c_dim 5 --batch_size 10 --log_step 5 --sample_dir stargan_celeba/samples --log_dir stargan_celeba/logs --model_save_dir stargan_celeba/models --result_dir stargan_celeba/results --selected_attrs Black_Hair Blond_Hair Brown_Hair Male Young

GPU measures

The following command will take 44 hours to complete on Colab (NVidia Tesla K80, compute capability: 3.7).

python main.py --mode train --dataset CelebA --image_size 128 --c_dim 5 --sample_dir stargan_celeba/samples --log_dir stargan_celeba/logs --model_save_dir stargan_celeba/models --result_dir stargan_celeba/results --selected_attrs Black_Hair Blond_Hair Brown_Hair Male Young

python main.py --mode=train --dataset Both --image_size 256 --c_dim 5 --c2_dim 8 --sample_dir stargan_both/samples --log_dir stargan_both/logs --model_save_dir stargan_both/models --result_dir stargan_both/results --batch_size 6 --log_step 5