samialabed / rlcache

Cache Manager using Reinforcement Learning
9 stars 5 forks source link

[Caching strategy]Reward rework #12

Open samialabed opened 5 years ago

samialabed commented 5 years ago

Currently, the reward is scaled based on hits and misses.

The agent isn't making use of the cache capacity available.

We should pass the cache capacity as an argument and scale the reward that it encourages caching at the beginning. i.e. reward is * (1 + cached_entry/cache_capacity)

Conduct experiments with this reward vs other rewards