Observers and Game Saves#357
Open
hopshackle wants to merge 44 commits into
Open
Conversation
…OBUST selection policies
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This has three main components:
Two advisers are provided. An Underdog adviser that will suggest better moves to players who are currently doing badly; and Tutor that is looking out for one specific player, and will suggest moves to them if they make one that the tutor thinks is particularly awful.
Addition of infrastructure to save a game state, and then restart from this saved state. This will always require game-specific methods in the game state to allow this to be saved to a JSON format and then resurrected from this. Backgammon (and its ancient variants) are implemented as an example in this PR, along with Exploding Kittens and Connect 4.
Facility has also been added to RunGames to allow a specific JSON state to be specified, and then start all games in the tournament from this specific state.
Some serialization of Components has be standardised to one of the two conventions used.
Addition of infrastructure to run a game, taking a snapshot of the GUI at each decision point, and of the MCTS search situation. This is encapsulated in a new MCTSDecisionRecorder listener that can be wired in via theRunGames config. This will also save the JSON state of each game state if IToJSON is supported.
It spits out a dot file of the MCTS tree at each point; anbd will also convert this to a png if Graphviz is locally installed.
Other changes
4) Connect4 now defaults to the standard 6 x 7 board (not 8x8, but this is now fully configurable)