Now, we would like to decode a feature vector with RNN$\text{'}$s$.$ The picture below illustrates how a vector encoding of the English sentence "I have seen better lectures" is translated into a sentence of a foreign language.

Unlike in encoding, at each step, an output distribution is produced in a decoding RNN$.$

Now, which of the following is true about decoding RNN$\text{'}$s$?$

$($Choose all those apply.$)$