[RLlib] - Algorithm.add_module
does not use the module_state
argument.
#46247
Labels
bug
Something that is supposed to be working; but isn't
P0
Issues that should be fixed in short order
rllib
RLlib related issues
rllib-multi-agent
An RLlib multi-agent related problem.
What happened + What you expected to happen
What happened
Using the
Algorithm.add_module
with amodule_state
does not use the module state, but instead loads or builds the module directly from the passed inSingleAgentRLModuleSpec
. This results in an error about missing network weights due to theinference-only
design.What you expected to happen
That a module state can be loaded into a module when calling
Algorithm.add_module
with theLearner
's module beinginference_only=False
and theEnvRunner
's module beinginference_only=True
. The module state should at best come from aLearner
's module (b/c it has all networks).Versions / Dependencies
Python 3.11
Ray master
Reproduction script
Issue Severity
Medium: It is a significant difficulty but I can work around it.
The text was updated successfully, but these errors were encountered: