/AI-Chess

AI chess

Primary LanguageJupyter Notebook

What is this

This is a AI based Chess Game

TO DO

  1. Make a playable game that choose a random move using python-chess: Done
  2. Use a NN to make the engine smart: Done
  3. Make a UI to play and test
  4. Improve the chess engine so that it can consistently beat me: Under Development
  5. Make the whole this class based, and remove python-chess from it.
  6. (maybe) Do step 3, but with Reinforcement Learning, RNNs, Transformers
  7. Beat Alphazero :)

CURRENT STAGE

The model is overfitting a bit. But it seems reasonably easy to make it fit better.

  • The model wouldn't know how to win. So make the normalisation of inp, out better
  • Make a ui to test better
  • Make the encoding better ?
  • Train a larger model