Based on PARL, the OAC algorithm of deep reinforcement learning has been reproduced, reaching the same level of indicators as the paper in Mujoco benchmarks.
Paper: OAC in Better Exploration with Optimistic Actor-Critic
Please see here to know more about Mujoco games.
- python3.5+
- parl>=2.0.0
- paddlepaddle>=2.0.0
- gym==0.9.1
- mujoco-py==0.5.7
# To train an agent for Humanoid-v1 game
python train.py
# To train for other game
python train.py --env [ENV_NAME]