In order to use Pytorch and applying basic deep learning examples we are going to use a pre-configured docker instance. This will help us getting a ready to use environment.
- To run Docker, your machine must have a 64-bit operating system running Windows 7 or higher.
- Make sure that Hyper-V is disabled if it is installed on your pc.
- If you are using Windows 10 you can directly download docker and install it from Docker Store.
- Otherwise, you can download and install Docker Toolbox.
- After installation is done, open new cmd window and type
docker ps. If you encountered an error running this command, you should follow these steps. - Finally, it is worth to spend few minutes on reading Get Started Tutorial
- Open new cmd window and run
docker image ls. It should list available docker images. - To install a Pytorch docker image we need first to download it. run
docker pull stepankuzmin/pytorch-notebook. - After downloading and unpacking is done, you can start new container by running this command
docker run -it --rm --name="PyT" -p 8888:8888 stepankuzmin/pytorch-notebook - Copy the link generated and paste it on new browser window. It should point to a jupyter notebook working directory.
- If link is not working try to replace
localhostby192.168.99.100 - Now you are ready to rock 👍
Does the PyTorch environment launched from Docker have access to the GPUs?