HOWTORUN.md

Setup and run

1. Set up WSL https://docs.microsoft.com/en-us/windows/wsl/install. Need to select Ubuntu last version
2. Enable NVIDIA CUDA on WSL https://docs.microsoft.com/en-us/windows/ai/directml/gpu-cuda-in-wsl. https://developer.nvidia.com/cuda-downloads
3. Set up nvidia docker https://github.com/NVIDIA/nvidia-docker
4. Download dataset, create dir shared on Ubuntu home user directore
5. Start docker service sudo service docker start
6. Run tensorflow container

sudo docker run -it --rm -p 8888:8888 angpysha/mytfgpu:1.0.6 -v ~/shared:/tmp/shared 7. Setup Pycharm or other internpreter to notebook server localhost:8888

How to enter bash to docker

1. Find container id using sudo docker container list
2. Run command sudo docker exec -it {containerId} /bin/bash

How to commit new changes to docker container

1. Find current running container ID using command sudo docker container list
2. Create new version of image sudo docker commit {id} angpysha/mytfgpu:{version}
3. Push chnages to docker repository sudo docker push angpysha/mytfgpu:{version}