Struct LSTMCell 

pub struct LSTMCell<B>
where
    B: Backend,
{
    pub input_size: usize,
    pub hidden_size: usize,
    /* private fields */
}

A single-step LSTM cell.

Shapes

• input x: [batch, input_size]
• hidden h: [batch, hidden_size]
• cell c: [batch, hidden_size]
• output (h’, c’): ([batch, hidden_size], [batch, hidden_size])

Fields

input_size: usize

hidden_size: usize

Implementations

impl<B> LSTMCell<B>

where B: Backend,

pub fn new( input_size: usize, hidden_size: usize, use_bias: bool, dtype: DType, device: &<B as Backend>::Device, ) -> Result<LSTMCell<B>, Error>

Create a new LSTMCell with Kaiming uniform initialization.

pub fn forward( &self, x: &Tensor<B>, h: &Tensor<B>, c: &Tensor<B>, ) -> Result<(Tensor<B>, Tensor<B>), Error>

Forward: compute (h’, c’) from (x, h, c)

• x: [batch, input_size]
• h: [batch, hidden_size]
• c: [batch, hidden_size]
• returns (h', c'): ([batch, hidden_size], [batch, hidden_size])

pub fn parameters(&self) -> Vec<Tensor<B>>

Return all trainable parameters.

pub fn named_parameters(&self) -> Vec<(String, Tensor<B>)>

Return all trainable parameters with names.