Dockerfile for machine learning environment(scikit-learn, chainer, gensim, tensorflow, jupyter)
Jupyter Notebook Python Dockerfile Shell
Switch branches/tags
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.

README.md

Purpose

This docker setting is for tring to touch and test some machine learning.

Installed main softwares

  • Tensorflow 1.10.0
  • Chainer 4.3.1
  • PyTorch 0.4.1
  • Keras 2.2.2
  • Chainer RL 0.4.0
  • Scikit-learn 0.19.2
  • Gensim 3.5.0
  • Numpy 1.14.5
  • Pandas 0.23.4
  • Jupyter Lab 0.33.10
  • Matplotlib 2.2.3
  • Mecab latest
  • Juman++ 7.01
  • NLTK 3.3.0

and other dependent libraries.

Password

Please update passwords(default is "ml" for following).

  • ml user password
  • ipyton password(jupyter_notebook_config.py)

How to run docker image

# Build image
# This image requires more than 13 GB disk space
docker build -t zuqqhi2/ml-python-sandbox .

# Run jupyter notebook & tensorboard
docker run -it -p 8888:8888 -p 6006:6006 zuqqhi2/ml-python-sandbox

# Login container
docker run -it -p 8888:8888 -p 6006:6006 zuqqhi2/ml-python-sandbox /bin/bash
source ~/.bash_profile
pipenv shell

# Set japanese locale
export LANG=ja_JP.UTF-8
export LC_ALL=ja_JP.UTF-8
export LC_CTYPE=ja_JP.UTF-8

Change Log from 1.0.3

  • 1.2.0
    • New Library/Tool : PyTorch, Chainer RL, Pipenv, Jupyterlab
    • Others : Change start_webuis.sh to run Jupyter Lab instead of Jupyter Notebook
  • 1.1.0
    • New Library/Tool : Seaborn, TFLearn, TFGraphviz, Tensorboard
    • Others : samples.ipynb to introduce how to use libraries, start_webuis.sh to run jupyter notebook and tensorboard

TODO

  • Nothing now

References