Skip to content

Latest commit

 

History

History
17 lines (15 loc) · 1.04 KB

DEVELOPMENT.md

File metadata and controls

17 lines (15 loc) · 1.04 KB

ReFinED Development

We recommend developing code on an EC2 instance with a GPU. We use p3.8xlarge as it has a large amount of memory and 4 GPUs which enables fast training.

  • Copy the ReFinED code to your machine git clone https://github.com/amazon-science/ReFinED.git.
  • Use your preferred IDE (such as PyCharm) to modify the code.
  • Copy the ReFinED code to your EC2 instance rsync -zarv --prune-empty-dirs --include "*/" --include "*.py" --exclude="*" ReFinED/ ubuntu@X.XXX.XXX.XX:/path_to_code/.
  • Add a new host to your ssh config file so that you do not need to provide the path to the private ssh key each time you connect or copy data (~/.ssh/config) example:
Host x.xxx.xxx.xxx
     User ubuntu
     IdentityFile <path_to_ssh_private_key_file>
  • SSH into your EC2 instance.
  • Install the Python dependencies using pip install -r requirements.txt.
  • Add ReFinED to your Python path export PYTHONPATH=$PYTHONPATH:/<path_to_code>/src/.
  • To confirm the setup is correct run python example_scripts/refined_demo.py runs successfully.