Code to setup AWS GPU instance to run Daniel Nouri's Facial Keypoints competition convoluted neural network code
To use this script, first run this to fit your first model: python kfkd.py fit Then train a bunch of specialists that intiliaze their weights from your first model: python kfkd.py fit_specialists net.pickle Plot their error curves: python kfkd.py plot_learning_curves net-specialists.pickle And finally make predictions to submit to Kaggle: python kfkd.py predict net-specialists.pickle
- start a AWS GPU instance
- I used Ubuntu Server 14.04 LTS (HVM), SSD Volume Type
- Search "community AMI's" for AMI ID
theano-tensorflow-cuda7.0-cudnn4 (ami-2f8c7b42)
- And for GPU instance I used GPU instances g2.2xlarge
- ssh into the server
git clone https://github.com/shawntsai/Facial-Keypoints-Detection.git
chmod 777 -R AWSGPU_DeepLearning/
vi cookies.txt
- Then copy-paste kaggle cookie into this file
./setup.sh