Earlier we were calculating the UCB for each arm index. I have changed it to use the previously calculated UCB, and update the UCB of only those arms that are pulled
before:
INFO:absl:Heuristic lilUCB's play with multi_armed_bandit runs 30.58 seconds.
after:
INFO:absl:Heuristic lilUCB's play with multi_armed_bandit runs 18.36 seconds.
Earlier we were calculating the UCB for each arm index. I have changed it to use the previously calculated UCB, and update the UCB of only those arms that are pulled
before: INFO:absl:Heuristic lilUCB's play with multi_armed_bandit runs 30.58 seconds. after: INFO:absl:Heuristic lilUCB's play with multi_armed_bandit runs 18.36 seconds.