explorative moves which are identical to best moves estimation will learn
This commit is contained in:
+1
-1
@@ -79,9 +79,9 @@ class MachinePlayer(APlayer):
|
||||
next_move = best_move
|
||||
is_exp = False
|
||||
if uniform() < self.p_exp:
|
||||
is_exp = True
|
||||
index = np.random.randint(len(moves))
|
||||
next_move = moves[index]
|
||||
is_exp = best_move != next_move
|
||||
elif do_sample:
|
||||
index = sample(values)
|
||||
next_move = moves[index]
|
||||
|
||||
Reference in New Issue
Block a user