open_spiel
open_spiel copied to clipboard
google-deepmind
OpenSpiel is a collection of environments and algorithms for research in general reinforcement learning and search/planning in games.
1. The `brew unlink` command appears to be failing now, so MacOS versions of Python on GitHub actions may have changed. Example of failed runs: https://github.com/deepmind/open_spiel/actions/runs/2873127141 (from commits in #894)...
@lanctot @ssokota: @gregdeon and I have adapted @vwxyzjn [CleanRL's](https://github.com/vwxyzjn/cleanrl) PPO implementation to work with OpenSpiel. In order to test the implementation, we have also wrapped Gym's Atari games into an...
As the question says, my code tries to compare two different game states. When is `state1 == state2`? From my limited knowledge of pybind11, I can see that equality for...
I get an error `bad_weak_ptr` when I try to use python games with C++ CFR. I'm 99% sure this used to work, but sometime over the past half-year it stopped...
Is it possible to release some well-trained models for the games supported by the open spiel framework such as connect four, breakthrough, and othello?
I get this error with CMake 3.16.3 > CMake 3.17 or higher is required. You are running version 3.16.3 In the installation guide, it still says 3.12 https://github.com/deepmind/open_spiel/blob/ae9371a1c92a6bfb2312e8315882bfe7d3e69a1a/docs/install.md?plain=1#L28
Hi all, I'm starting a thread to collect requests for new games and plans for upcoming games. We will keep this updated as new games get released, and as requests...
It looks like ExtensiveToMatrixGame defaults to creating a payoff matrix of all pure strategies but that is far from desirable now. Especially with the new sparsification techniques, is there any...
In open_spiel/python/example/deep_cfr, the tabular method is usually used to solve the exploitability or NashConv of kuhn poker. """ average_policy = policy.tabular_policy_from_callable(game, deep_cfr_solver.action_probabilities) conv = exploitability.nash_conv(game, average_policy) """ Can we find...
Metadata
Owner
Metadata
OpenSpiel is a collection of environments and algorithms for research in general reinforcement learning and search/planning in games.