I don't know how the evaluation of the training data should be done

[ml6362662]

After manually training 20 episodes using pvp/experiments/metadrive/train_pvp_metadrive.py, a file will be generated in runs/pvp_metadrive_toy_gamepad_gamepad containing data, models, pvp_metadrive_toy_gamepad_gamepad_ 1 and monitor.csv these files, I don't know how to evaluate the training data. Is it using a pvp/eval_script/metadrive/eval_pvp_td3_metadrive.py file for evaluation?,But I don't know how to set the command line parameters.,If anyone knows how to evaluate it, I'd like to let me know in detail.,Thank you very much.