Determining the correct number of neurons for a neural network

I do some research with neural networks, and the concept and theory as a whole make sense to me. Although one question that suits me, to which I have not yet been able to find an answer, is how many neurons should be used in a neural network. to achieve appropriate / effective results. Including hidden layers, neurons on a hidden layer, etc. Do more neurons increase more accurate results (with a greater load on the system) or are fewer neurons still sufficient? Is there any correct rule that can help determine these numbers? It depends on the type of training / learning algorithm that is embedded in the neural network. Does it depend on the type of data / input that is displayed on the network?

If it will be easier for you to answer questions, I will most likely use the functions of forward and backpropogation as the main method of training and forecasting.

On the side of the note, is there a prediction / firing rule algorithm or a training algorithm that is usually regressed as “best / most practical”, or does it also depend on the type of data presented on the network?

Thanks to anyone who has any data, it is always appreciated!

EDIT: As for the C # tag, this is the language in which I will assemble my neural network. If this information helps at all.