I'd like to ask how to play the trained strategy to check its performance

[KevinXFG]

I didn't see the file play.py,Thank you.

[maedmatt]

Hey @KevinXFG , there's no play.py in order to evaluate a trained policy, use eval_agent.py:

python src/holosoma/holosoma/eval_agent.py  --checkpoint=<path_to_checkpoint.pt>

The checkpoint can be a local .pt file or a W&B URI (wandb:////<run_id>/<checkpoint_name>). See training README for details.