David J Wu
David J Wu
Sorry for missing this earlier. However, I think I don't understand your question, possibly due to a poor translation to English. What is the problem, specifically? Does it work with...
KataGo's model size switches were accomplished by concurrently training the next model on the same data as the smaller network, and then choosing to switch at roughly the point that...
Thanks for reporting, I'll take a look.
Thanks for the report. Looks like pytorch 2.6 was released this year, I'll see about incorporating this fix.
Cool, sounds like a fun project. It sounds like from your message you might not have seen yet the "analysis engine" option in KataGo? The main doc on that is...
I'm hoping to get a release out before too long with some minor bugfixes and such, and want to do some minor experiments with search features that could add small...
Even on weak hardware, it may often be the case that the *largest* neural network is the best, due to being so much stronger than the faster and smaller networks...
Simple page is now up! See here https://katagotraining.org/extra_networks/
Thanks, will fix.
Is this in katrain or something like that? Unfortunately these parameters will only work for GTP-based guis.