DSIT Docker

Provides a Notebook environment for DSIT on Docker (include CUDA environment)
Some google colab specific features cannot be used.

Based Docker image

• TensorFlow Jupyter
  https://hub.docker.com/r/tensorflow/tensorflow/tags?page=1&name=jupyter

Added Modules

Python Modules

• pandas
• seaborn
• scikit-learn
• graphviz
• xgboost
• lightgbm
• imblearn
• pydot
• opencv-python
• tqdm
• imageio
• scikit-image
• statsmodels
• jupyterlab
• jupyterlab-language-pack-ja-JP

Jupyter Extensions

• @lckr/jupyterlab_variableinspector
• @jupyterlab/git

How to use

1. Set the version of TensorFlow you want to use to $TF_PACKAGE_VERSION
   If $TF_PACKAGE_VERSION is blank, it will be 'latest'.
   ex.)
   2.20.0 (<- TensorFlow 2.20.0)
   2.20.0-gpu (<- TensorFlow 2.20.0 and CUDA)
   latest (<- latest version (default))
   latest-gpu(<- latest version and CUDA)

2. Build image and run container, then jupyter lab server will start.

3. Access the following URL with a web browser.
   http://localhost:8888/lab?token=dsit

Build image example

# set package version
export TF_PACKAGE_VERSION="2.20.0-gpu"
# build image
docker build -t ghcr.io/tbtech-section4/dsit:${TF_PACKAGE_VERSION} --build-arg TF_PACKAGE_VERSION=${TF_PACKAGE_VERSION} ./

# run container
docker run --rm -it --gpus all -p 8888:8888 -v ${PWD}/Notebook:/tf/Notebook -t ghcr.io/tbtech-section4/dsit:${TF_PACKAGE_VERSION}