search

Jamesflynn1 / CS344-Opponent-Exploitation-Poker

A third year uni project aiming to implement and evaluate the EFR algorithm with different deviation types and explore a potential tradeoff between exploitability and expected value of a strategy in practice.
0 stars 0 forks source link

Verification of Correct algorithm #27

Closed Jamesflynn1 closed 1 year ago

Jamesflynn1 commented 1 year ago

Current informed deviations are is the action deviation module.

Jamesflynn1 commented 1 year ago

Misunderstanding, tested informed cf and blind cf

Jamesflynn1 commented 1 year ago

Blind CF matches almost exactly with CFR with no linear averaging, no CFR plus regret reset and no alternating updates. Repeat runs * 3 and perform 1000 iterations.

Jamesflynn1 commented 1 year ago

Works for CFR with minor overflow issues with larger numbers, blocked by #31

Jamesflynn1 commented 1 year ago

Happy enough to move this into a final report issue.