You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Typically the inputs are modified with a linear layer which is omitted here for flexibility (it would look like $\mathbf{x}^{n}\_t = \mathbf{W}^{n} \mathbf{x'}\_t$). This operation can be parallelized along the sequence dimension in contrary to the recurrent part, \\
0 commit comments