KataGo attempts to "solve" 7x7


This week, KataGo finished a project to try to “solve” 7x7. Uploaded here are comprehensive browsable opening books of all the believed-optimal lines. Also some commentary on some cool new variations and moves missed by earlier human attempts to solve 7x7. I also highly recommend reading these earlier attempts if you like this kind of thing, some of the earlier articles are really nice to give a high-level summary of the lines even if they might be superseded in some details now (unfortunately for the later Chinese solution attempts, or Google translate and OCR translation isn’t the best).

These aren’t solutions in the sense of rigorous mathematical solutions, of course, they’re just an attempt to get as close to perfect as we can - and who knows, if we’re lucky, maybe we actually reached it now. :slight_smile:

Also, KataGo is still improving at https://katagotraining.org/, any spare GPU power or contributors could still help. Especially recently things have been very slow and additional contributors over the next months could make a big difference so that we may begin to try to upgrade the net size to 60b. If you like supporting this research or KataGo generally, help would be greatly appreciated! Or if you know people or organizations that might be interested to help, let me know.

I hope you find these 7x7 books interesting! (also posted on reddit and a couple other places)


Happy to get a third chance to like/upvote this! The work you’re doing with KataGo goes above and beyond what we could only dream about a few years back.


Very interesting. But I’m wondering about 6x6 and smaller: which board sizes have been solved mathematically, and where are the optimal solutions published?



More about solutions for other small board sizes