Really bad neural network, but it works it has 3 inputs and one output, with no hidden layers mutation rate should be around 0.5 for quickest learning i would say change the mu. rate to 0.2 after the number on the bottom is bigger than 1000 the number on the bottom is the best distance + the average distance / 2 which the best possible number is 2000 (perfect in both directions all you have to do is press play and watch for turbo mode -> shift + green flag press r to reset the ai (randomize weights/biases) there are other maps as backrounds set mutation rate to 0 and press green flag to see the best ai that you trained
