Are there important advances from Graves and others' earlier work on differentiable memory (associative LSTM, neural turing machines, neural GPUs etc.)? Or is t

Are there important advances from Graves and others' earlier work on differentiable memory (associative LSTM, neural turing machines, neural GPUs etc.)? Or is this just that getting re-reported in Nature for prestige?