: thoroughly-tested. In many cases,
we verified against known values and/or reproduced results from papers.
~: implemented but lightly tested.
| Game | Reference | Status |
|---|---|---|
| Backgammon | Wikipedia | |
| Breakthrough | Wikipedia | |
| (Uncontested) Bridge bidding | Wikipedia | |
| Catch | Recurrent Models of Visual Attention, Behaviour Suite for Reinforcement Learning, Appendix A |
~ |
| Coin Game | https://arxiv.org/abs/1802.09640 | ~ |
| Connect Four | Wikipedia | |
| Cooperative Box-Pushing | https://arxiv.org/abs/1206.5295 | ~ |
| Chess | Wikipedia | |
| First-price Sealed-bid Auction |
Wikipedia | |
| Go | Wikipedia | |
| Goofspiel | Wikipedia | |
| Havannah | Wikipedia | |
| Hex | Wikipedia | ~ |
| Kuhn poker | Wikipedia | |
| Leduc poker | Bayes’ bluff: Opponent modelling in poker | |
| Liar's Dice | Wikipedia | |
| Markov Soccer | https://arxiv.org/abs/1609.05559 | ~ |
| Matching Pennies (three-player) |
"Three problems in learning mixed-strategy Nash equilibria" | |
| Oshi-Zumo | "Solving the oshi-zumo game" http://mlanctot.info/files/papers/aij-2psimmove.pdf |
|
| Oware | Wikipedia | |
| Pentago | Wikipedia | |
| Phantom Tic-Tac-Toe | Following this PhD thesis | ~ |
| Pig | Wikipedia | |
| Tic-Tac-Toe | Wikipedia | |
| Tiny Bridge | |
|
| Y | Wikipedia | |