Struct LSTM 

pub struct LSTM { /* private fields */ }

LSTM Layer - processes entire sequences

Implementations

impl LSTM

pub fn new( input_size: usize, hidden_size: usize, num_layers: usize, bidirectional: bool, dropout: f32, ) -> Self

Create a new LSTM layer

Arguments

• input_size - Size of input features
• hidden_size - Size of hidden state
• num_layers - Number of stacked LSTM layers
• bidirectional - Whether to use bidirectional LSTM
• dropout - Dropout probability between layers

pub fn forward_sequence(&self, input: &Tensor) -> Tensor

Forward pass through LSTM

Arguments

• input - Input tensor of shape [batch, seq_len, input_size]

Returns

Output tensor of shape [batch, seq_len, hidden_size * num_directions]

Trait Implementations

impl Module for LSTM

fn forward(&self, input: &Tensor) -> Tensor

Forward pass

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

Get all trainable parameters

fn train(&mut self)

Set module to training mode

fn eval(&mut self)

Set module to evaluation mode

fn is_training(&self) -> bool

Check if in training mode

fn num_parameters(&self) -> usize

Number of trainable parameters