training an agent to move around and play the game according to some reward funciton is pretty easy, but having an AI that can figure out the strategies and direction to finish a regular ddnet map is very hard.
My tasing algorithm could maybe finish a short 2 player map if given 1-4 weeks but anything above 2 players is completely out of the question.
likely far too much work especially since it would take the same verification as a namechange but i wonder if they could have a list of people who volunteer to publicize their teehistorian files (and if things like chat are logged maybe have that stripped)
02:54
but im not even sure how teehistorian works its probably on a per map/server basis so there might just not be any easy way to separate specific players
training an agent to move around and play the game according to some reward funciton is pretty easy, but having an AI that can figure out the strategies and direction to finish a regular ddnet map is very hard.
My tasing algorithm could maybe finish a short 2 player map if given 1-4 weeks but anything above 2 players is completely out of the question.
it's like making a chess ai by training on millions of human games. Of course it will work but it won't be better than the humans its trained on, nor will it beat chess engines that just brute force billions of positions.
I have a question that interests me very much: how did you get my map? xd
because I didn’t seem to give any download links, and the only person I gave it to seems to have no connection with you? I'm just wondering how you got it, I don't mind you having it.
RedRay32
I have a question that interests me very much: how did you get my map? xd
because I didn’t seem to give any download links, and the only person I gave it to seems to have no connection with you? I'm just wondering how you got it, I don't mind you having it.
1
1
8
