Examples of config.yaml for supported games
natravedrova opened this issue · 8 comments
Would you mind sharing sample settings for all games supported by your excellent framework?
There is config.yaml for TicTacToe already. It would be great to have similar files for other games (config.geister.yaml, config.geese.yaml or commented section in the existing config). I am not sure who are your target audience, but as a newcomer, not all parameters might be obvious.
Any comments in the files would be much appreciated as well.
Thank you for your suggestion and your trial. I agree your advice about target audience.
We are talking about preparing the configs for each game.
And we're planning to describe the parameters and the features of HandyRL in README.
e.g.
- What is
forward_stepsparameter? - HandyRL enables policy gradient with off-policy correction, self-play, losses (V-Trace, UPGO)...
What information would make it easier for you and beginners to use?
Please wait for a while...
Thanks
What information would make it easier for you and beginners to use?
Thank you @ikki407! Your plan described in the comment look great.
New PR for parameters. I added more details for training.
#117
And we will make config for each game in another PR.
@natravedrova
I added the parameter descriptions https://github.com/DeNA/HandyRL/blob/master/docs/parameters.md.
This was merged and the readme updated.
Can you see that?
Thanks
@natravedrova
I added the parameter descriptions https://github.com/DeNA/HandyRL/blob/master/docs/parameters.md.
This was merged and the readme updated.
Can you see that?
Thanks
This is fantastic! Thank you very much!
This includes a lot of technical terms. So the explanation how to tune or set the parameters might be noted.
I will do that after some time...
Hi @natravedrova, we shared the parameters used to train a model for hungry geese in Kaggle discussion. Please check it out.
https://www.kaggle.com/c/hungry-geese/discussion/218190
I am completely satisfied with all the updates related to config.yaml :-)