RNN and Language modeling
·
AI/이론
RNN and Language modeling¶ Basic structure¶ An unrolled recurrent neural network¶ Input and outputs of RNNs(rolled version)¶ $h_{t-1}$: old hidden-state vector $x_{t}$: input vector at some time step $h_{t}$: new hidden-state vector $f_{W}$: RNN function with parameters W $y_{t}$: output vector at time step t (can be obtained through $h_{t}$) ❗ 여기서 주의 점은 매 time step마다 같은 함수와 같은 parameters를 사용한다...