search

quantumiracle / nash-dqn

Official code of Nash-DQN for paper: Nash-DQN algorithm for two-player zero-sum Markov games, details see our paper: A Deep Reinforcement Learning Approach for Finding Non-Exploitable Strategies in Two-Player Atari Games. Zihan Ding, Dijia Su, Qinghua Liu, Chi Jin
17 stars 2 forks source link

general-sum nash #5

Closed hwz9612 closed 1 year ago

hwz9612 commented 1 year ago

Hi, Recently, I solve the general-sum nash equilibrium by modify the code. But I meet with one problem. How can I modify the update function? When we solve zero-sum nash equilibrium, we only construct a network and get a q-values. When we solve general-sum nash equilibrium, do I construct two network?

quantumiracle commented 1 year ago

Hi,

Yes, I think it should either has two networks or a shared network body with two heads.