over 3 years ago

Update to default board size, availability of training data.

Hi all!

As some of you may be aware, the example games of Hex that were listed on our repository are currently unavailable since the site is down. To compensate, we have made some example games available on the repository here. Note that the format of the training data is described in this README. This data will be updated as more becomes available, with ~10,000 training games likely to be available by Wednesday. These games are from self-play by the high-level program “Wolve”, set to a mid-range search depth. Fun fact: Wolve was developed here at the University of Alberta!

Unfortunately, the timely creation of the above training data has forced us to change the size of the board to be used in the tournament. We will now be playing games on a 10×10 board rather than an 11×11 board. This is primarily to ensure the training data is generated quickly enough for all who want to use supervised learning to get started.

Questions?

If you have any questions about the hackathon, please post on our discord in #hex-tournament, or on the discussion forum.