Login to hydra and start an interactive session.
ssh hpc.nyu.edu
ssh hydra <
qsub -q interactive -I -l nodes=hydra1.es.its.nyu.edu:ppn=1:gpus=1,walltime=4:00:00,mem=10gb
Clone repository (if you have not done so already). Make sure to replace login with your own git login.
git clone https://login@github.com/wojzaremba/hydra.git
Rename old ~/.local directory and add symboic link to repository_loc/.local (where repository_loc is the location of the hydra repository):
cd ~/
ln -s repository_loc/.local
Add the following to your ~/.bash_profile
export PYTHONPATH=$PYTHONPATH:repository_loc/local/lib
export PATH=repository_loc/local/bin:$PATH
module load python/2.7.5
module load atlas/3.10.1
Installation : sh build.sh
Execution :
python convnet.py --data-path=./data/cifar-10-py-colmajor/ --save-path=./checkpoints --test-range=6 --train-range=1-5 --layer-def=./example-layers/layers-80sec.cfg --layer-params=./example-layers/layer-params-80sec.cfg --data-provider=cifar --test-freq=10 --epochs=8