MyGo

It's my Go bot!

Experimental Go bot implemented in PyTorch.

PS: This project has nothing to do with MyGO!!!!!, except its name.

References

• Mainly the book: Deep Learning and the Game of Go, Chinese version named 《深度学习与围棋》.
• The repos: BetaGo, Code and other material for the book.
• The papers:
  • AlphaGo: Mastering the game of Go with deep neural networks and tree search.
  • AlphaGo Zero: Mastering the game of Go without human knowledge.
• GNU Go: the CLI design reference.
• SGF file format FF[4].
• GTP - Go Text Protocol.
• Tromp-Taylor rules.
• Monte Carlo tree search - Wikipedia.

Notes

• Chapter 7 training reproducibility.
• PyTorch CrossEntropyLoss expects the input to contain the unnormalized logits, which is different from Keras.