Struct RnnConfig 

pub struct RnnConfig {
    pub d_input: usize,
    pub d_hidden: usize,
    pub bias: bool,
    pub initializer: Initializer,
    pub batch_first: bool,
    pub reverse: bool,
    pub clip: Option<f64>,
    pub hidden_activation: ActivationConfig,
}

Configuration to create a Rnn module using the init function.

Fields

d_input: usize

The size of the input features.

d_hidden: usize

The size of the hidden state.

bias: bool

If a bias should be applied during the Rnn transformation.

initializer: Initializer

Rnn initializer

batch_first: bool

If true, the input tensor is expected to be [batch_size, seq_length, input_size]. If false, the input tensor is expected to be [seq_length, batch_size, input_size].

reverse: bool

If true, process the sequence in reverse order. This is useful for implementing reverse-direction RNNs (e.g., ONNX reverse direction).

clip: Option<f64>

Optional hidden state clip threshold. If provided, hidden state values are clipped to the range [-clip, +clip] after each timestep. This can help prevent exploding values during inference.

hidden_activation: ActivationConfig

Activation function applied to the hidden state before computing hidden output. Default is Tanh, which is standard for Rnn.

Implementations

impl RnnConfig

pub fn new(d_input: usize, d_hidden: usize, bias: bool) -> RnnConfig

Create a new instance of the config.

Arguments

Required Arguments

d_input

The size of the input features.

d_hidden

The size of the hidden state.

bias

If a bias should be applied during the Rnn transformation.

Optional Arguments

clip

Optional hidden state clip threshold. If provided, hidden state values are clipped to the range [-clip, +clip] after each timestep. This can help prevent exploding values during inference.

• Defaults to None

Default Arguments

initializer

Rnn initializer

• Defaults to "Initializer::XavierNormal{gain:1.0}"

batch_first

If true, the input tensor is expected to be [batch_size, seq_length, input_size]. If false, the input tensor is expected to be [seq_length, batch_size, input_size].

• Defaults to true

reverse

If true, process the sequence in reverse order. This is useful for implementing reverse-direction RNNs (e.g., ONNX reverse direction).

• Defaults to false

hidden_activation

Activation function applied to the hidden state before computing hidden output. Default is Tanh, which is standard for Rnn.

• Defaults to "ActivationConfig::Tanh"

impl RnnConfig

pub fn with_initializer(self, initializer: Initializer) -> RnnConfig

Sets the value for the field initializer.

Rnn initializer

• Defaults to "Initializer::XavierNormal{gain:1.0}"

pub fn with_batch_first(self, batch_first: bool) -> RnnConfig

Sets the value for the field batch_first.

If true, the input tensor is expected to be [batch_size, seq_length, input_size]. If false, the input tensor is expected to be [seq_length, batch_size, input_size].

• Defaults to true

pub fn with_reverse(self, reverse: bool) -> RnnConfig

Sets the value for the field reverse.

If true, process the sequence in reverse order. This is useful for implementing reverse-direction RNNs (e.g., ONNX reverse direction).

• Defaults to false

pub fn with_hidden_activation( self, hidden_activation: ActivationConfig, ) -> RnnConfig

Sets the value for the field hidden_activation.

Activation function applied to the hidden state before computing hidden output. Default is Tanh, which is standard for Rnn.

• Defaults to "ActivationConfig::Tanh"

pub fn with_clip(self, clip: Option<f64>) -> RnnConfig

Sets the value for the field clip.

Optional hidden state clip threshold. If provided, hidden state values are clipped to the range [-clip, +clip] after each timestep. This can help prevent exploding values during inference.

• Defaults to None

impl RnnConfig

pub fn init<B>(&self, device: &<B as BackendTypes>::Device) -> Rnn<B>

where B: Backend,

Initialize a new Rnn module.

Trait Implementations

impl Clone for RnnConfig

fn clone(&self) -> RnnConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Config for RnnConfig

fn save<P>(&self, file: P) -> Result<(), Error>

where P: AsRef<Path>,

Saves the configuration to a file.

fn load<P>(file: P) -> Result<Self, ConfigError>

where P: AsRef<Path>,

Loads the configuration from a file.

fn load_binary(data: &[u8]) -> Result<Self, ConfigError>

Loads the configuration from a binary buffer.

impl Debug for RnnConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for RnnConfig

fn deserialize<D>( deserializer: D, ) -> Result<RnnConfig, <D as Deserializer<'de>>::Error>

where D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Display for RnnConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl Serialize for RnnConfig

fn serialize<S>( &self, serializer: S, ) -> Result<<S as Serializer>::Ok, <S as Serializer>::Error>

where S: Serializer,

Serialize this value into the given Serde serializer.