rickstaa/stable-learning-control

Clean up SAC/LAC code

Closed this issue · 2 comments

User story

As a developer I want to have clean code that we can ship to the rest of the group.

Considerations

  • Create class out of SAC and LAC implementations.
  • Create a clear one line run command for both the inference and the training.
    • For this we can use a command like python machine_learning_control.py train.py --alg=SAC --env=Oscillator --run=run_124839 and machine_learning_control.py inference.py --alg=SAC --env=Oscillator --run=run_124839`. In this when no run number is given it automatically uses the last trained run of a given environment.

Acceptance criteria

  • Clean run command.
  • Clean clear code.
  • Automatic last run option.

sure

This is fixed in d571487