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