large-scale-curiosity
large-scale-curiosity copied to clipboard
How can I load the trained model and watch the performance of the agent?
I run the run.py and it can work, the default environment is Breakout, and the model seems to be stored in 'env0_0.pk'. I want to know how to load the trained model and watch the performance of the agent in the graphical interface.
I have the same question! help
'env0_0.pk' is not a model file. It has information(selected actions, rewards, first random state) of one episode. I think it is to reproduce that episode. And there is no codes loading/saving a model in the repo. You need to add it.