Early Nash-Q Research Notes¶

Historical link collection from before the repository's native tabular Nash-Q implementation. See PettingZoo Nash-Q for the current design.

Tabular Nash-Q Learning¶

• OpenSpiel seems best, also has NashConv, so
• https://github.com/karthikv792/NashQ/tree/main - bugged and for 2 agents only

Deep Nash-Q Learning¶

• https://github.com/p-casgrain/Nash-DQN
• capable of 2+, only tested with 2, https://github.com/ismorphism/DeepGame