KaTrain v1.0 released - play against various AI options, instant feedback and review

A while back I made a tool with the idea of using AI to get instant feedback on your moves, finding reviewing with lizzie and such to be not quite what I needed.
This led to KaTrain, a tool which uses KataGo’s scoreLead function to give you a ranking of your move by the point loss they cause, but without showing the best moves in advance. With the time available due to recent events, I managed to rewrite this such that the program is ready for a wider audience.

The new release is available here for windows or here for source code and manual.


  • Review your games to find the moves that were most costly in terms of points lost.
  • Play against AI and get immediate feedback on mistakes with option to retry.
  • Play against a wide range of weakened versions of AI with various styles.
  • Automatically generate focused SGF reviews which show your biggest mistakes, in a format optimized for sabaki.
  • View best moves according to analysis as well as neural network policy.
  • Support for non-square boards, up to 52x52 with the appropriate katago version.

And much more!

The AI options have also been deployed on OGS as bots for the last few weeks, although in the program you have more power to configure them. The ‘weighted’ AI is particularly recommended for kyu players and appears reasonable in its mistakes.

Enjoy, and let me know what you think! Suggestions are very welcome.


This looks very interesting. May I start with a suggestion.
The tuning phase happens in the console but the interface doesn’t show any hint that something is going on. So I was trying various buttons, confused that nothing was happening, before checking what was happening in the console. Maybe a message “Tuning in progress, please wait for a while” would help.

Thanks for the suggestion. :slight_smile: I put it in the manual, but starting to realize manuals are a bit 1995. It’s surprisingly hard to detect this tuning, but I will probably make a first startup welcome message which mentions this.

The thing is - reading the manual, I know some tuning is going to take place at some point, but I am not sure when. If I had just the console, or if the interface was greyed out, I would understand I need to wait. Here, I have access to an interface, no special message and the buttons work, so I assume the tuning will start when the analysis begin or something?

It’s when katago first starts, made an issue here: https://github.com/sanderland/katrain/issues/26
Likely to be fixed in the next version one way or another. Thanks again for the suggestion.

This is now improved in the new version v1.0.4

1 Like