Practical Deep Learning For Coders

Setup

1. ** Create amazon AWS account **

2. ** Request EC2 limit upgrade **

   • https://aws.amazon.com/contact-us/ec2-request
   • Ireland (EU)
   • p2.xlarge
   • reason: fast.ai MOOC

3. ** Download Anaconda for python 2.7 **

   • https://www.continuum.io/downloads

4. ** Install AWS CLI **

   • pip install awscli
   • check by typing aws. If it complains about locale's check this site

5. ** Create new AWS user **

   • https://console.aws.amazon.com/iam/home?region=eu-west-1#/users$new?step=details
   • [username]
   • Check 'Programmatic access'
   • next
   • Add policies directly -> 'AdministratorAccess'
   • next
   • Copy / paste Access key ID & Secret access key somewhere safe (credentials.txt)

6. Configure AWS

   • aws configure
   • Enter previous generated credentials
   • Region: eu-west-1
   • Output: text

7. Download setup script

   • Check the git repo https://github.com/fastai/courses/tree/master/setup
   • Download setup file:

   wget https://raw.githubusercontent.com/fastai/courses/master/setup/setup_p2.sh

   • chmod 775 setup_p2.sh

8. Run setup

   • ./setup_p2.sh
   • Copy output in credentials
   • Copy connection command

9. Connect to server

   • ssh -i /Users/[localUser]/.ssh/aws-key.pem ubuntu@[instanceUrl]
   • When prompted type 'yes'

10. Update packages & reboot

    • Undo bug with history: sudo rm .bash_history
    • sudo apt-get update
    • sudo apt-get upgrade
    • sudo reboot now

11. Check setup

    • After reboot, login again and check nvidia drivers: nvidia-smi
    • Trouble shoot http://wiki.fast.ai/index.php/AWS_install

12. Start notebook

    • jupyter notebook
    • Goto url: http://[instanceUrl]:8888
    • Default pass 'dl_course'

13. Check notebooks

    • Create new notebook in 'nbs' folder usding Python [conda root]
    • check basic command (like 1+1)
    • check imports import theano & import keras

14. Shortcuts

    • Download aliases wget https://raw.githubusercontent.com/fastai/courses/master/setup/aws-alias.sh
    • source aws-alias.sh

15. Setup notebooks

    • Fork the gitrepo https://github.com/fastai/courses

    Perhaps rename it to something sensible, like fastai-courses?

    • SSH to machine (aws-ssh)
    • cd ~ if you are not already there
    • Clone your repo git clone https://github.com/[repoUrl]
    • Make a link to notebooks ln -s ~/[repoName]/deeplearning1/nbs/ nbs/deeplearning1
    • Run jupyter in background jupyter notebook &> /dev/null &
    • Make data dir mkdir ~/nbs/data and change to it cd ~/nbs/data
    • Download dogs&cats wget http://www.platform.ai/data/dogscats.zip
    • Install unzip sudo apt-get install unzip
    • Unzip the data unzip -q dogscats.zip
    • Go to your notebook http://[instanceUrl]:8888/notebooks/deeplearning1/lesson1.ipynb
    • Step through the process of lesson1.ipynb

16. Kaggle CLI

    • Create profile at https://www.kaggle.com/
    • On server: pip install kaggle-cli
    • kg config -g -u [username] -p [password] -c dogs-vs-cats-redux-kernels-edition
    • mkdir ~/nbs/data/dogscats.2 and cd to it
    • Go to https://www.kaggle.com/c/dogs-vs-cats-redux-kernels-edition and accept competition requirements (by downloading the data)
    • On server: kg download