Open ShowiBin opened 1 year ago
Hi This code does not train the agent for any maze without further training. But I now that it is possible. I suggest that you check out "meta learning". This paper can be insightful: http://proceedings.mlr.press/v70/finn17a/finn17a.pdf
thanks for your help,
you are welcome
I'm working on train a agent able to run any maze map without further training. Is this code finished this? Is this task impossible?