Check out blog posts about this here: https://andrewkho.github.io/wordle-solver/
Env setup
python3 -m venv server-env
source server-env/bin/activate
pip install -r requirements.txt
Node setup
nvm install 17.8.0 # node version
npm install
To update requirements
pip install -r server-requires.txt
pip freeze > requirements.txt
# The first line of requirements.txt needs to be
# -f https://download.pytorch.org/whl/torch_stable.html
# and then the torch==1.8.1+cpu in order to keep the
# heroku slug-size below 500mb
Local dev
# Start server
gunicorn --pythonpath deep_rl app:app
# Start react dev server
npm run start-local
For pre-deploy local testing
npm run build-local # Build static site
heroku local # Site should be available at localhost:5000
Local testing looks for the pre-trained model at data/checkpoints/a2c_deployed.ckpt so create a symlink there or something.
To deploy
git push heroku master # Deploys site
To deploy new Lightning model/checkpoint
./deploy_checkpoint.sh <path_to_checkpoint>