You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
weiji14
changed the title
Setup reproducible scientific coding environment using docker
Setup reproducible scientific coding/development environment using docker
Jun 22, 2017
Install build-essential and python3-dev within mouthful docker RUN statement in dockerfile to build latest jupyter notebook (et al. packages) using python3-pip.
Install hydrogen using apm-beta, initiate jupyter notebook with default config for default atom user and implement a hack fix for the 'No module named ipykernel_launcher' to do with using pure python3.
Mostly copy-and-pasted iPython implementation of the neural network from scratch tutorial by Denny Britz. Edited to be Python3 compatible (e.g. use print() instead of print, and range instead of xrange).
Also added scientific libraries to atom-beta dockerfile by pip3 installing standard data science modules. Namely matplotlib, numpy, scikit-learn and scipy.
Initialize with a 'stable' debian 10 buster base (said as of late Jun 2017), and setup python with a few useful libraries.
See e.g. https://doi.org/10.1038/546173a on the importance of reproducibility
The text was updated successfully, but these errors were encountered: