David J Wu

Thanks for the report. It's somehow surprisingly hard to find any comprehensive documentation on nvidia's website about the compatibility matrix between different versions of CUDA and different GPUs, but after...

For an overview of self-play training.sh, see https://github.com/lightvector/KataGo/blob/master/SelfplayTraining.md. See in particular: > You probably want to read the comments and edit the parameters in python/selfplay/synchronous_loop.sh and in the selfplay and...

Glad you're interested - see https://github.com/lightvector/KataGo/blob/master/SelfplayTraining.md and take a look at the script https://github.com/lightvector/KataGo/blob/master/python/selfplay/synchronous_loop.sh. You will probably want to edit this script and the config files it points to yourself...

For setting the rules for training, see https://github.com/lightvector/KataGo/blob/master/cpp/configs/training/selfplay1.cfg#L81 and https://lightvector.github.io/KataGo/rules.html

Your question is hard for me to understand and doesn't give many details - what are you trying to do? Can you give more details?

Thanks for the feedback! It just so happens that I've been working on 9x9 for a little while now, I've been preparing to maybe add a 9x9 to [katagobooks.org](https://katagobooks.org/), which...

Cool, thanks for the report! Would you be able also to share the SGF? I would be interested in taking a look at it too.

If it's just the neural network and getting GPU acceleration libraries and such to work that's the problem but you have the ability to run other processes like GnuGo, then...

What happens if you manually open the GTP console and type in "komi 4.5" or whatever other number you want to change the komi to and hit enter?

(there is an explicit manual GTP console within lizzie)