Tensorflow exam can have different version requirements for the exam.
Here is how you can change a few numbers and run couple commands to get you container setup.
Overall you need to build your image, push to docker hub and use it in paperspace gradient.
- Clone the repository.
- Install
docker
in your system. https://www.docker.com/products/docker-desktop/ - Change the version number in this line of Dockerfile
FROM tensorflow/tensorflow:2.9.0-gpu-jupyter
- Replace 2.9.0 with a number required by you exam eg:
FROM tensorflow/tensorflow:2.10.0-gpu-jupyter
- Make sure
tensorflow/tensorflow:2.10.0-gpu-jupyter
image is present in docker hub. - To verify type go to https://hub.docker.com/r/tensorflow/tensorflow/tags type
eg: 2.10.0
and see if there is an image present. - Change
RUN pip install
packages to the reqired version. - Do not remove jupyterlab package in the end.
- Go inside the colned project dir and run
docker build --tag my_image .
- Create an account on docker hub and push your local image.
- Go to gradient and use your image(in the advanced section of notebooks) from docker hub like
docker-hub-username/dokcer-image-name-you-used-for-docker-hub
. - you are all set.