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

cleaned up

Browse Source
This commit is contained in:
jens
2024-06-15 12:27:10 +02:00
parent 93dbeadd6d
commit 054c9b849c
1 changed files with 2 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
k-bandits.py
+2 -4
View File
@@ -14,10 +14,8 @@ def bandit(bias):
if __name__ == '__main__': if __name__ == '__main__':
kv = 1.0 epsilon = 0.0 # Anti-greediness
epsilon = 0.1
rho = 0.01 # reduce epsilon with age rho = 0.01 # reduce epsilon with age
alpha = 1.0
qu = np.zeros(K) qu = np.zeros(K)
nu = np.zeros(K) nu = np.zeros(K)
@@ -43,7 +41,7 @@ if __name__ == '__main__':
# get reward from bandit # get reward from bandit
r = bandit(ql_star[a]) r = bandit(ql_star[a])
nu[a] = nu[a] + 1 nu[a] = nu[a] + 1
qu[a] = qu[a] + alpha*(r - qu[a])/nu[a] qu[a] = qu[a] + (r - qu[a])/nu[a]
r_sum += r r_sum += r
r_vec.append(r_sum/N) r_vec.append(r_sum/N)
# 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)