Resources - kirillbobyrev/pabi Wiki
Here is a list of useful resources to draw the ideas from. Also, see a list of chess-related repositories on GitHub.
- Pruning Game Tree by Rollouts
- Thinking Fast and Slow with Deep Learning and Tree Search
- Watch the Unobserved: A Simple Approach to Parallelizing Monte Carlo Tree Search
- Temporal difference learning and TD-Gammon
- TD-Gammon, a self-teaching backgammon program, achieves master-level play
- A Rollout-Based Search Algorithm Unifying MCTS and Alpha-Beta
- A Survey of Monte Carlo Tree Search Methods
- Learning to Search with MCTSnets
- Batch Monte Carlo Tree Search
- Monte Carlo Tree Search: A Review of Recent Modifications and Applications
- On Effective Parallelization of Monte Carlo Tree Search
- The Second Type of Uncertainty in Monte Carlo Tree Search
- Dual Monte Carlo Tree Search
Engines with papers and code
Popular and influential ones
- Leela Chess 0 (lc0) (as an Open Source version of Alpha Go Zero)
- Crazy Ara
- Slow Chess
- shakmaty is not a chess engine but an implantation of most important routines in Rust. A good reference to compare my imlpementations to.
- Walleye is a medium-strength chess engine in Rust with a good implementation.
- Very fast move generator in Rust
These environments can be used to estimate Pabi performance (ELO) and see it play agains other engines.
- lc0 self-play games training data, e.g. this one on GDrive: taken from Stockfish commit from lc0 blog post
- Chess Engine in C video series on YouTube