Pytorch Lightning Implementation of the Dreamer-RL.
| Deepmind Control Suite Environment | GIF | Avg Reward while testing |
|---|---|---|
| Walker - Walk | Each episode contains 1000 steps, per episode reward = avg reward per step * 1000 |
|
| Acrobot - Swingup |
Dreamer - Paper by Danijar Hafner, Timothy Lillicrap, Jimmy Ba, Mohammad Norouzi