CS-7641-assignments
CS-7641-assignments copied to clipboard
Adds multiprocessing support for q_learning experiment.
This adds a process pool for executing the q_learning parameter search in parallel. I found that the process pool performed better than a thread pool which is why I went that route.
This is pretty crude, but I wanted to get it up so others can look at it and use it if they want to speed up a4 runs.
@"Jeremy Brown" in ml class if anyone has questions.
Wrong @jeremy btw.
On Mon, Apr 8, 2019 at 18:25 Farga83 [email protected] wrote:
@jeremy https://github.com/jeremy Brown if anyone has questions.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/cmaron/CS-7641-assignments/pull/20#issuecomment-481065746, or mute the thread https://github.com/notifications/unsubscribe-auth/AAAAx4-EIN_Q1BZkOHKDbKtpjSWyc14jks5ve-wWgaJpZM4cjZ-x .
sorry, should have quoted that. Didn't realize it separated first/last in the mention.
This could cause more harm than good for most folks. It will use all your ram due to q-learner stats collection if you are not careful. For that reason, I'd probably not merge it in as is, but perhaps keep the pr open for folks inclined to work through any issues they find.
Only tested on Ubuntu 18.04 native, 32 GB ram, python 3.6.8