Browse Teach Curate

LSTM Output Gate

A sigmoid gate that filters the tanh of the cell state to produce the hidden state exposed to the next layer and time step. It decouples what the cell remembers internally from what it reveals, so long-term memory need not leak into every prediction.

jeremy Video

Long Short-Term Memory LSTM Architecture and Training in Recurrent Neural Networks

Long Short-Term Memory (LSTM) networks address the vanishing and exploding gradient problems inherent in vanilla Recurrent Neural Networks by architecting a distinct cell state path for long-term mem…