#2656 - Corrected from_config() for ActionPenalty so that it can pull the negative reward value from YAML and apply, defaulting to 0 still if not found/not configured.
Currerntly prints to terminal when a negative reward is being applied, though this is for implementation and troubleshooting. To be removed before PR is pushed out of draft
This commit is contained in:
@@ -739,6 +739,11 @@ agents:
|
||||
options:
|
||||
agent_name: client_2_green_user
|
||||
|
||||
- type: ACTION_PENALTY
|
||||
weight: 1.0
|
||||
options:
|
||||
agent_name: defender
|
||||
penalty_value: -1
|
||||
|
||||
agent_settings:
|
||||
flatten_obs: true
|
||||
|
||||
Reference in New Issue
Block a user