Hi,
I have some questions on the observation and action space.
The observation space in the colab for PianoWithShadowHands
has a goal
key. What exactly is that?
![Screen Shot 2023-05-25 at 4 24 51 PM](https://user-images.githubusercontent.com/13341926/241048099-d09185d8-4deb-4e75-8863-4a5f87b8b484.png)
Next, could you give some details about the reduced action space? Is it helpful for learning? Does it limit the agent's final performance?