-
python packages
- pytorch>=0.4.0
- torchvision>=0.2.1
- tensorboardX
-
PPO codes are borrowed from the course assignment4 https://github.com/cuhkrlcourse/ierg6130-assignment/tree/master/assignment4
python train.py --algo PPO --log-dir data/PPO_gamma_1 --input-length 10
Tensorboard visualization:
tensorboard --logdir=data/PPO_gamma_1/runs/ --port={port_num}
Note that all the experiments above will save the tensorboard log file in data/PPO_gamma_1/runs/ directory