ValueError: Could not get observation and action spaces from remote worker. Maybe specify them manually in the config?

RLlib Configure Algorithm, Training, Evaluation, Scaling
1

I get the error ValueError: Could not get observation and action spaces from remote worker. Maybe specify them manually in the config? when I try to run some rllib code in an ubuntu vm (I think this is important as the vm might have different network settings)

The same code runs fine on my desktop. How can I configure RLLib properly so it works on my ubuntu VM?

1 Like
2

I have the same issue.
Does anybody has a good link where explained algorithm configuring?

3

Hi @Nexon33 and welcome to the discussion forum,

it is hard to say what i shappening there without a working example of code. On my Fedora 37 System I do not face any problems - nor do I face them on my Ubuntu clusters on GCP.