TeviotDataScienceGame

This repository is for the edinburgh university data science game entry.

Installation Guide (With OpenBLAS)

OpenBLAS/numpy install

There is a script provided to use all of ashburys cores:

ssh student.compute
chmod a+x install.sh
./install.sh
export OMP_NUM_THREADS=40
. cord3/core3/bin/acitivate

This creates a venv valled core3 inside a folder called cord3.

Project install

After installing openblass/numpy install this project such that the imports work do the following command on DICE machines (within the root directory of this project)

# inside virtual env locally with BLAS numpy preinstals
pip install -e .
# or globally in dice but without BLAS/Numpy core config facility
pip install --user -e .

All this within the virtualenv you are now ready to train the current keras models in train_scripts using all of ashburies cores. The only things you need to do when logging in are activating the venv . core3/bin/acitivate and setting the enviroment variableOMP_NUM_THREADS=40 .