Was really looking forward to your game as you used neural network as well !
Unfortunately, I am not fond of card games, so it is hard for me to get into your entry :/
In short, the AI seemed to behave quite nicely, so this is very nice.
I can really imagine the amount of time you spent just training the model !
What indicators did you use to see if they were indeed improving in the game ?
Did you used advsersarial stratefy when they had to compete against each other ?
And for the cheater part, did you used a different model for each cheat ability, or did you have one big model for all opponents and just switched off some outputs ?
Viewing post in Cheater's Inn jam comments
Yeah, I settled on a card game early on so that I could spend most of my time on training the AIs instead of on game design. (Which isn't normally a good idea but that's what I wanted to get out of this jam.) I had ten groups of neural networks play thousands of games against eachother each round, replacing the worst scoring ones with mutations and combinations of the better ones, and repeating that for thousands of rounds. Each cheat ability has its own pool of neural network because I wanted them to learn to use it to their advantage. For example the Spy can always see all players' cards and I wanted its neural network to be trained to use that information. But structurally all NNs have the same shape, the main difference lies in their starting hand; so one cheater always starts with a Twelve of Hearts, but other NNs can learn to pick it up if it's on the table.
The code I wrote for it is open source, if your interested: https://github.com/SLiV9/MegaNaiadsTrainer