Implement many Sparse Reward algorithms in Gym Fetch environment
We implemented many classes of Sparse Reward algorithms in Gym Fetch environment including Reward Shaping, Imitation Learning, Curriculum Learning, Hindsight Experience Replay, Curiosity-Driven Exploration, Hierachical Reinforcement Learning. This work is for better understanding of sparse reward algorithms.
Our code is based on https://github.com/andrew-j-levy/Hierarchical-Actor-Critc-HAC- and we have changed a lot on code simplification and content richness.
python main.py --retrain
python main.py --retrain --rtype dense
python main.py --retrain --curriculum 2
python main.py --retrain --imitation --imit_ratio 1
python main.py --retrain --her
python main.py --retrain --curiosity
python main.py --retrain --layers 2
python main.py --test
if using HDDPG, you should use :
python main.py --test --layers 2
python main.py --retrain --her --save_experience