Heya folks, as I said in the title, Katrain is exceedingly mean to me. I’m about as beginner as beginner gets, and I’d like to be able to configure Katrain to be less, well, mean.
So far, the most I can find about actually configuring Katrain to be “easier” is some reddit posts, some GitHub issues, and this goldmine that I am, unfortunately, far too out of my depth to read. (I did try.)
I was hoping someone could point me to some resources on the matter. Even just a YouTube video with a good rundown of how to set it super low (the lowest I’ve seen it behave is 15k, which is laughably out of my league) would help me start to builda mental model of how to hold this.
Obligatory apologies if this is the wrong category for this.
# Available options are:
# preaz_{RANK from 20k to 9d} - imitate player of given rank, before AlphaZero opening style became popular
# rank_{RANK from 20k to 9d} - imitate player of given rank, after human openings changed due to AlphaZero.
It’s still technically out of my league, but not by nearly as much, and this’ll probably give me a better idea how I should play than a traditional bot that’s been lobotomized.
If you mean because they’re more user friendly, meh. I’m pretty comfortable with arcane software (I am a long-time developer, a ham radio operator, and user of chess engines; I don’t mind the arcane). I’d rather use the thing that’s more featured, even if it’s hard to use.
If you mean because they’re easier to beat, fair.
Hahaha.
My issue here isn’t so much the tool is hard to use it’s that I don’t know anything about configuring go engines at all.
What I meant by pre AI is about the quality of the moves, not the UI design.
AI based software have still difficult time to play at low level, I mean to be interesting in offering their challenge. In some way pre AI are more elaborated and fit better to a low level public.
Now I still strongly recommend to play with humans as with bots, play with low level humans like you, it’s really much more interesting.
At best take a go set with you and teach someone who could spend some time playing together.
Sorry, I forgot to @xela in the last message, that was directed at them not you.
RE: Humans vs bots, that’s a very fair point, and one I’d prefer. I have something like 50 correspondence games going on OGS as we speak. Unfortunately, my wife has thus far bounced off of Go, but I hold out hope that my oldest son (3) will pick it up like he has chess (I have recently gotten him to stop simply naming and then throwing the pieces!)
I just also like playing against engines and learning how they work; my chess game improved dramatically when I wrote my own chess bot, and I’m doing the same for Go (slowly, over time). I want to make sure I know how to configure an existing engine because it’s a great way to get insights into what your own bot is doing wrong!
Really, for me, playing against and making bots for games is almost a wholly separate hobby from playing the same game with other people.
I mean that (some of) the older engines play weak but still coherent moves, with a more natural (dare I say human-like?) flow to the game, making them easier to learn from. If you “cripple” a strong AI to bring it down to 10 kyu level, some of its moves will be just random nonsense.
Having said that, I’ve not yet tried the latest KataGo upgrade: maybe that does something better on a low rank setting?
Not to necro my pwn post, but I finally sat down today to look at this, and I’m not sure if I’m doing it right.
When I open up katrain now, katago boots up just fine in the background and even says it has access to the human mode, but when I play against katago, it still says it’s 9d.
I have a feeling it’s just katrain not knowing what to do with the “human” configuration stuff yet, but if anyone knows better I’d love to hear.
