Struct TransformerDecoderBlock 

pub struct TransformerDecoderBlock {
    pub attention: CausalSelfAttention,
    pub ln1: LayerNorm,
    pub ffn: FeedForward,
    pub ln2: LayerNorm,
}

Transformer decoder block (GPT-style with causal attention).

Fields

attention: CausalSelfAttention

Causal self-attention

ln1: LayerNorm

First layer norm

ffn: FeedForward

Feed-forward network

ln2: LayerNorm

Second layer norm

Implementations

impl TransformerDecoderBlock

pub fn new( n_embd: usize, n_head: usize, max_seq_len: usize, dropout: f32, layer_norm_eps: f32, activation: &str, ) -> Self

Creates a new transformer decoder block.

Trait Implementations

impl Debug for TransformerDecoderBlock

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

Formats the value using the given formatter.

impl Module for TransformerDecoderBlock

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

Performs the forward pass.

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

Returns all parameters of this module.

fn train(&mut self)

Sets the module to training mode.

fn eval(&mut self)

Sets the module to evaluation mode.

fn named_parameters(&self) -> HashMap<String, Parameter>

Returns named parameters of this module.

fn num_parameters(&self) -> usize

Returns the number of trainable parameters.

fn set_training(&mut self, _training: bool)

Sets the training mode.

fn is_training(&self) -> bool

Returns whether the module is in training mode.

fn zero_grad(&self)

Zeros all gradients of parameters.

fn name(&self) -> &'static str

Returns the module name for debugging.