To make all our experiment results reproducible, we submit codes with hyperparameters used for our method on each task in this supplementary material.

Please download the pretrained model weights from google drive:
https://drive.google.com/drive/folders/1IYLEiS6uJh8nvz0vlrvroVhUGcCDm27W?usp=sharing

and put them in the folder "model_weights"

Then you can test these pretrained model on each task by running scripts:

python test_PPO_pick.py --config '../config/test_pick.yaml'

or if you want to train model yourself, run:

python train_PPO_pick.py --config '../config/train_pick.yaml'