self-play: 155 parameters not match with the called function

[guohl]

No description provided.

[Narsil]

Which version are you running ? This should be fixed in master.

[guohl]

A___line: 155(self_play.py): game_data = play_game(model, model, mcts_simulations, conf['STOP_EXPLORATION'], self_play=True, resign_model1=resign, resign_model2=resign)
B___line: 32(self_play.py): def play_game(model1, model2, mcts_simulations, stop_exploration, self_play=False, num_moves=None, resign_model1=None, resign_model2=None)

[guohl]

I changed A to "game_data = play_game(model, model, mcts_simulations, conf['STOP_EXPLORATION'], True, None, resign_model1=resign, resign_model2=resign) and the result seems correct! and the game record seems reasonable.

[Narsil]

You should not have non named arguments (None is non named argument, after self_play=True, which is a named argument). Which version of python are you runninng ? Also num_moves should be None by default

[guohl]

the latest master

[guohl]

"game_data = play_game(model, model, mcts_simulations, conf['STOP_EXPLORATION'], True, None, resign_model1=resign, resign_model2=resign"

as above

[Narsil]

Which version of python are you running ?

[guohl]

python 3.5

[guohl]

it works fine only produces to much moves!
if add self fill eyes check to mask in play.py(line:12) it might be much better

[guohl]

[Narsil]

Ok, I'm closing this, for the fill eyes use #12 plz.

[guohl]

a few hours, it seems reasonable!

[guohl]

ok, thank you

[Narsil]

Don't be so sure, I've been running it on 9x9 for quite a while, it's still easily beatable. Looking at leela it looks like it needs quite a few games before making decent progress.