always calc nu
This commit is contained in:
+1
-1
@@ -78,10 +78,10 @@ def test(_k_arms: int, _episode_len: int, ql_star, _tie_break, _param: Param) ->
|
||||
_reward = _reward_z[j] + ql_star[_a]
|
||||
|
||||
# calc
|
||||
_nu[_a] = _nu[_a] + 1
|
||||
if _param.alpha > 0:
|
||||
_qu[_a] = _qu[_a] + _param.alpha * (_reward - _qu[_a])
|
||||
else:
|
||||
_nu[_a] = _nu[_a] + 1
|
||||
_qu[_a] = _qu[_a] + (_reward - _qu[_a]) / _nu[_a]
|
||||
|
||||
# Reduce tendency to explore with number of steps (or with age for humans)
|
||||
|
||||
Reference in New Issue
Block a user