jens/k-bandits
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

refactored

Browse Source
This commit is contained in:
jens
2025-04-04 16:36:07 +02:00
parent 5f4c554c17
commit e5603da9a9
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
k-bandits.py
+3 -3
View File
@@ -78,11 +78,11 @@ def test(_k_arms: int, _episode_len: int, ql_star, _tie_break, _param: Param) ->
_reward = _reward_z[j] + ql_star[_a] _reward = _reward_z[j] + ql_star[_a]
# calc # calc
_nu[_a] = _nu[_a] + 1 _nu[_a] += 1
if _param.alpha > 0: if _param.alpha > 0:
_qu[_a] = _qu[_a] + _param.alpha * (_reward - _qu[_a]) _qu[_a] += _param.alpha * (_reward - _qu[_a])
else: else:
_qu[_a] = _qu[_a] + (_reward - _qu[_a]) / _nu[_a] _qu[_a] += (_reward - _qu[_a]) / _nu[_a]
# Reduce tendency to explore with number of steps (or with age for humans) # Reduce tendency to explore with number of steps (or with age for humans)
_epsilon = _param.epsilon * (1 - _param.rho) _epsilon = _param.epsilon * (1 - _param.rho)