Finally generated a checkpoint, how do I see use the checkpoint now to see if the training succeeded?

[jaganrvce1]

• Is there a sample code someone can point me to that I can point me to checkpoint to see if the training worked?
• There seems to be two checkpoints, one is the critic and actor? How do I combine them in a single program?

I would appreciate if someone points me to a piece of code that shows how actor/critic are used together etc. Thanks!

[prvnsmpth]

I think you can just load up the actor model using huggingface transformers and run inference. The critic model only exists to evaluate the actor model's outputs during training.