Topic: Test problem for

Good evening! There is a neural network which does not have specific target - it is necessary to look generally as it works. The method of reverse propagation of an error is used.
Cannot throw any test problem, where some neurons in an input layer (<=5), a little in hidden (<=5) and one on an output.
It is necessary that was some dial-up of learning vectors, each of which contained values of the inputs, a demanded output and starting values of weight factors for all neurons.
If know a similar problem, tell, please