Real-valued Attractor Network implementation

Is there a Matlab implementation of an attractor network that can handle real-valued data?

Hopfield networks exist in the Neural Network Toolbox but can only learn patterns where the values are either -1 or +1.