-
Notifications
You must be signed in to change notification settings - Fork 13
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
save and restore env state before evaluation
- Loading branch information
1 parent
895837b
commit cf60925
Showing
1 changed file
with
2 additions
and
0 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
cf60925
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Seems like it would be better to use the functions from RLInterface rather than accessing the
state
field. What if someone tries to use this with a model that conforms to RLInterface, but doesn't have astate
field, like a ROS simulator or something? I would recommend resetting the environment if it is done.cf60925
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks, you are perfectly right, unfortunately there is nothing in the interface right now that would support saving the current state of the environment and restoring it later.
The bug encountered is due to the fact that the evaluation and exploration are using the same
env
object, hence during evaluation the potential hidden state inenv
are changed inconsistently with the current state ofenv
in the exploration.I think about the following options:
env
, one for exploration and one for evaluationhidden_states
andset_hidden_states
to RLInterface.I will file an issue in DeepQLearning.jl and will try to address this soon