GlobalEval #28
GlobalEval #28
Conversation
bowen-xu
requested review from
maxeeem
and removed request for
MoonWalker1997
|
@maxeeem could you please help to review the code when you get a chance? |
|
Locations where the for evaluators are updated: |
|
@bowen-xu there seems to be a conflict with the dev branch which will need to be resolved before merging |
| task, concept, self.memory) | ||
|
|
||
| # update well-being | ||
| self.global_eval.update_wellbeing(task_executed.truth.e) |
There was a problem hiding this comment.
This is currently inside the mental_operation function. Is it just the name or is there another function elsewhere for other types of operations?
The reason I ask is if well-being is meant to represent the 'body' functioning as expected, then it seems more related to operations like ^move or ^look etc rather than mental operations.
There was a problem hiding this comment.
Oh u r right. Since there is no "body" in the current implementation, Pei told me that we can tentatively update well-being when an operation is succesfully executed. So this should also be called when any operations (like ^move and ^look) are executed.
| ''' Here, belief is not None, and it is the best solution for the task | ||
| Thus, do global evaluation to update satisfaction of the system. | ||
| ''' | ||
| self.global_eval.update_satisfaction(task.achieving_level(belief.truth), task.budget.priority) |
There was a problem hiding this comment.
We should probably add this to questions as well in addition to goals. But perhaps later, not in this initial PR.
pynars/NARS/Control/Reasoner.py
Outdated
| We tentatively exploit the truth of a revised task to indicate alertness | ||
| """ | ||
| if judgement_revised is not None: | ||
| self.global_eval.update_alertness(judgement_revised.truth.c - task.truth.c) |
There was a problem hiding this comment.
Should this be an absolute value to avoid negative numbers?
There was a problem hiding this comment.
judgement_revised.truth.c - task.truth.c could be negative
There was a problem hiding this comment.
Should be OK, since the Revision rule always increases confidence, the Task's Judgment would always be lower confidence than the revised Judgment
To solve issue https://github.com/bowen-xu/PyNARS/issues/8