brianprichardson

Yes. You can reduce the number of simulations here: https://github.com/Zeta36/chess-alpha-zero/blob/master/src/chess_zero/config.py#L15 self.simulation_num_per_move = 1200 to a smaller value and it should play moves faster, but the moves won't be as "good".

Nobody knows the answers to any of those questions...experiment.

In situations like this you could simply use an older version. For instance, I am stuck on tensorflow 1.5 due to my old CPU.

python-chess (0.22.1) is the version I use, but I have also modified it slightly for supervised learning to try to parse both UCI and SAN move formats. I think those...

It has been a while, but for a "self" play run try without any weight file and it should create one to start with. The best weights are for "uci"...

Depending on the situiation, the weights (.h5) and model (.json) files must match the net architecture in the configs file (typically mini.py). The stronger ones that I uploaded do not...

First only do: python src/chess_zero/run.py uci Then, after it loads enter: uci [wait for uciok] isready [wait for readyok] go [should see some bestmove output but may take some time...

See #75 there is a link a fork with a working version.

Try run.py self again, but this time do not have any best model files. This will make it create a new model and start self play training from there.

If there is no model_best_config.json file, the error should not be the same. It should say something like no best model found, creating... Also, there is a lot of output...