How severe does this issue affect your experience of using Ray?
- Medium: It contributes to significant difficulty to complete my task, but I can work around it.
Gymnasium allows to define Discrete spaces starting at negative numbers, eg, Discrete(3, start=-1) would contain -1, 0, 1.
I want to use such a discrete space with negative values as action space and have configured my environment accordingly.
Can Ray RLlib handle such action spaces with negative actions? After all the model outputs logits as an array for all possible actions, where arrays are always zero-indexed. What happens with the negative actions then? It seems like they are never selected by the model.