RNN Components

$3$ points possible $($graded$)$

The main challenge with an $n-$gram model is that history needs to be variable, not fixed. Which parts of the

RNN allows for this?

$($Choose all that apply.$)$

The input layer which takes in new information and the previous state

Having a hidden state

The output layer specifying a probability distribution

Which aspect of the RNN differentiates it from a traditional feedforward neural network?

The hidden state is fed in as input for the next step

Uses nonlinear activation functions, such as softmax

Architecture transforms the previous layer with a weight matrix and adds a bias element

Is the following sentence true or false: The hidden state at step t only contains information about words close

to t $.$

True

False