docker link

1. https://hub.docker.com/repository/docker/x37n/python_dl/general
2. (automated build) https://hub.docker.com/repository/docker/x37n/python_dl/general

• This dockerfile aims to zsh, load anaconda, python, tensorflow, pytorch, several statistical library, and jupyterlab.
• From this repository, dockerHub automatically build latest image. Current latest file is built under the versions below
  • os: ubuntu XX.04
  • cuda, cudnn*-devel
  • PyTorch, TorchVision
  • Tensorflow
  • Jax / Flax

1. Build docker image based on Dockerfile. You can edit docker image name in this file.

bash build.sh

2. Create docker container based on the created docker image

start.sh

In container, it calls zsh as default. You can open juputerlab by

jupyter lab --ip=0.0.0.0 --port=8888 --allow-root

and you can see the logs at tensorboard by

tensorboard --logdir=<your logging directory>

3. Once you generated docker image, you can save it to share different servers.

docker save hitoshi/cuda11.2:2023.05.04 > 20230504_cuda112.tar

From tar file, you can load it without rebuilding.

docker load < 20230504_cuda112.tar

You can rename image tag by the following command line.

docker image tag hitoshi/python-dl:10.1-ubuntu18.04-F22T17 hitoshi/python-dl:10.1-ubuntu18.04-F22T17-v2

to update image(hitoshi/allinone:latest) from container (hitoshi_all)

docker commit hitoshi_all hitoshi/allinone:latest

You can also see backlogs for old Dockerfiles. Good Luck 😉