Hi, in GRUCellEx, the code: `hy = newgate + inputgate * (hidden - newgate)` But in the paper: you wrote: h_i_(t+1) = (1-u_i_(t)) * q_i_(t) + u_i_(t) * h_i_(t) Could you please help me explain this difference? Thank you so much
Hi, in GRUCellEx, the code:
hy = newgate + inputgate * (hidden - newgate)But in the paper: you wrote:
h_i_(t+1) = (1-u_i_(t)) * q_i_(t) + u_i_(t) * h_i_(t)
Could you please help me explain this difference? Thank you so much