Michael Pang
Michael Pang
Do you have files in your play_data folder? Although I agree the optimizer should check for that first.
Ohh okay, as it stands currently you have to run "sl" on them first to convert into json, then "opt". But if it's helpful I have a version in [my...
I just pushed again, but it might be easier to just use sl from this repo until I'm done experimenting with stuff.
Good point, I tried using a [custom evaluation function](https://github.com/Zeta36/chess-alpha-zero/blob/master/src/chess_zero/env/chess_env.py#L143) too but it might not be enough.
@ddugovic I'm not sure I understand what you mean. Are we predicting the elo of the players? Or are you talking about self-play evaluator, in which case what would the...
Does it help to change [start method](https://docs.python.org/3.6/library/multiprocessing.html#contexts-and-start-methods) to spawn?
Sorry I couldn't test it, I only have a windows laptop. Please let us know if you figure it out! On another note: without tensorflow-gpu how much cpu % does...
I have a modest GPU (GT 750M) so a total of about one game per minute on 3 processes. Interestingly my CPU usage is only around 30-40%.
Thanks for the encouragement! The tactics is a worry but at least it's fun to play with (very aggressive lol). What do you think of the idea of adding "knight-shaped"...
Sure, go for it! :) I'll continue experimenting with 5x5 convolutions, tell me if you have any thoughts on this.