I used tune.run to train my custom environment and had automatic checkpoints created during the process.
Now I want to load these checkpoints and analyze them via the
render() method of my environment to see how the agents behave at this point of time of the training.
I restore the model via
trainer.restore(checkpoint_path), but how do I execute the render method of my environment?
I could run a single more training step by calling
trainer.train() and then cause a render with
config["render_env"] = True, but is there another way?
Thanks in advance!