New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

More in-depth training statistics #1

Open

3 of 7 tasks

ed1d1a8d opened this issue Feb 26, 2022 · 0 comments

Collaborator

ed1d1a8d commented Feb 26, 2022 •

edited

Loading

It would be helpful to get more informative statistics during selfplay/victimplay training.

Useful statistics to track:

Win rate
Fraction of games won by resignation
Win-margin
Number of moves taken

It would be nice to be able to plot all of the above as a function of training epoch, with the following filters:

Color
Board size
Whether game was won or lost

Checked items means we already have support for that feature in some capacity.

Some comments:

It would be nice to code this up in a way that supports both training and match play (i.e. by calling the match command in KataGo).
The notebooks in https://github.com/HumanCompatibleAI/go_attack/tree/main/notebooks may contain some useful code for parsing statistics from games.

The text was updated successfully, but these errors were encountered:

ed1d1a8d mentioned this issue

More in-depth training statistics AlignmentResearch/KataGo-custom#4

Closed

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment