Post #1376969

@ramin_hal9001 Thanks for your memory. I am increasingly thinking that sharing and sharing between particular programmers' original neural networks is important. The reason I say that the equation I am using is an implementation of deep learning, I specifically mean that it is equivalent to a conventional feedforward neural network of a single hidden layer (with odd activation function) - Krotov and Hopfield, 2016. It is possible to choose an "attention" update function as well (Ramsauer 2020)