- a variation of LSTM. It’s more efficient cause it combines the hidden state with the cell state (into just one hidden state)
- unlike LSTM, it also has a
1-operator, so when it tells the hidden state to forget information, it also writes to it with new info.
- unlike LSTM, it also has a