Deploying RL agents to the cloud – a trading Bot-as-a-Service
The ultimate goal of training an RL agent is to use it for taking actions given new observations. In the case of our stock trading SAC agent, we have so far learned to train, evaluate, and package the best performing agent model to build our trading bot. While we focused on one particular application (autonomous trading bot), you can see how easy it is to change the training environment or agent algorithms based on the recipes in earlier chapters of this book. This recipe will walk you through the steps to deploy the Docker containerized/packaged RL agent to the cloud and run the Bot-as-a-Service.
Getting ready
To complete this recipe, you will need access to a cloud service such as Azure, AWS, GCP, Heroku or another cloud service provider that allows you to host and run your Docker container. If you are a student, you can make use of GitHub’s student developer pack (https://education.github.com/pack...
