encoder

encoder

Bidirectional transformer encoder for sequence encoding.

TransformerEncoder

TransformerEncoder(number_of_layers, embedding_dimension, number_of_heads, number_of_key_value_heads=None, feedforward_dimension=None, dropout=0.1, attention_dropout=0.0, activation=value, normalization_type=value, attention_type=value, positional_encoding_type=None, maximum_sequence_length=2048, bias=True, normalization_epsilon=1e-06, initializer_range=0.02)

Bases: TransformerMixin, Module

Bidirectional transformer encoder for sequence encoding.

Processes all tokens in parallel with bidirectional self-attention.

Initialize transformer encoder.

Parameters:

Name	Type	Description	Default
number_of_layers	int	Number of encoder layers.	required
embedding_dimension	int	Model embedding dimension.	required
number_of_heads	int	Number of attention heads.	required
number_of_key_value_heads	int | None	Number of K/V heads (for GQA).	None
feedforward_dimension	int | None	FFN hidden dimension.	None
dropout	float	Dropout probability for residual connections.	0.1
attention_dropout	float	Dropout probability for attention weights.	0.0
activation	str	Activation function (use ActivationFunction enum values).	value
normalization_type	str	Type of normalization (use NormalizationType enum values).	value
attention_type	str	Type of attention (use AttentionType enum values).	value
positional_encoding_type	str | None	Type of positional encoding (or None).	None
maximum_sequence_length	int	Maximum sequence length for positional encoding.	2048
bias	bool	Whether to use bias in linear layers.	True
normalization_epsilon	float	Epsilon for normalization layers.	1e-06
initializer_range	float	Standard deviation for weight initialization.	0.02

Source code in src/versatil/models/layers/transformer/encoder.py

def __init__(
    self,
    number_of_layers: int,
    embedding_dimension: int,
    number_of_heads: int,
    number_of_key_value_heads: int | None = None,
    feedforward_dimension: int | None = None,
    dropout: float = 0.1,
    attention_dropout: float = 0.0,
    activation: str = ActivationFunction.SWIGLU.value,
    normalization_type: str = NormalizationType.RMS_NORM.value,
    attention_type: str = AttentionType.GROUPED_QUERY.value,
    positional_encoding_type: str | None = None,
    maximum_sequence_length: int = 2048,
    bias: bool = True,
    normalization_epsilon: float = 1e-6,
    initializer_range: float = 0.02,
):
    """Initialize transformer encoder.

    Args:
        number_of_layers: Number of encoder layers.
        embedding_dimension: Model embedding dimension.
        number_of_heads: Number of attention heads.
        number_of_key_value_heads: Number of K/V heads (for GQA).
        feedforward_dimension: FFN hidden dimension.
        dropout: Dropout probability for residual connections.
        attention_dropout: Dropout probability for attention weights.
        activation: Activation function (use ActivationFunction enum values).
        normalization_type: Type of normalization (use NormalizationType enum values).
        attention_type: Type of attention (use AttentionType enum values).
        positional_encoding_type: Type of positional encoding (or None).
        maximum_sequence_length: Maximum sequence length for positional encoding.
        bias: Whether to use bias in linear layers.
        normalization_epsilon: Epsilon for normalization layers.
        initializer_range: Standard deviation for weight initialization.
    """
    super().__init__()
    self.number_of_layers = number_of_layers
    self.embedding_dimension = embedding_dimension
    self.number_of_heads = number_of_heads
    self.maximum_sequence_length = maximum_sequence_length
    self.initializer_range = initializer_range
    self.number_of_residual_blocks = 2  # Self-Attention + Feedforward
    self.number_of_key_value_heads, self.head_dimension = (
        self._resolve_attention_dimensions(
            embedding_dimension=embedding_dimension,
            number_of_heads=number_of_heads,
            number_of_key_value_heads=number_of_key_value_heads,
            attention_type=attention_type,
        )
    )
    self._setup_positional_encoding(
        positional_encoding_type=positional_encoding_type,
        embedding_dimension=embedding_dimension,
        maximum_sequence_length=maximum_sequence_length,
        number_of_heads=number_of_heads,
    )
    self.layers = nn.ModuleList(
        [
            TransformerEncoderLayer(
                embedding_dimension=embedding_dimension,
                number_of_heads=number_of_heads,
                number_of_key_value_heads=number_of_key_value_heads,
                feedforward_dimension=feedforward_dimension,
                dropout=dropout,
                attention_dropout=attention_dropout,
                activation=activation,
                normalization_type=normalization_type,
                attention_type=attention_type,
                bias=bias,
                normalization_epsilon=normalization_epsilon,
            )
            for _ in range(number_of_layers)
        ]
    )
    self.final_normalization = create_normalization_layer(
        normalization_type=normalization_type,
        dimension=embedding_dimension,
        epsilon=normalization_epsilon,
    )
    self.apply(self._init_weights)

forward

forward(hidden_states, padding_mask=None)

Forward pass through transformer encoder.

Parameters:

Name	Type	Description	Default
hidden_states	Tensor	Input embeddings (B, seq_length, D).	required
padding_mask	Tensor | None	Optional padding mask (B, seq_length).	None

Returns:

Type	Description
Tensor	Output hidden states (B, seq_length, D).

Source code in src/versatil/models/layers/transformer/encoder.py

def forward(
    self,
    hidden_states: torch.Tensor,
    padding_mask: torch.Tensor | None = None,
) -> torch.Tensor:
    """Forward pass through transformer encoder.

    Args:
        hidden_states: Input embeddings (B, seq_length, D).
        padding_mask: Optional padding mask (B, seq_length).

    Returns:
        Output hidden states (B, seq_length, D).
    """
    sequence_length = hidden_states.shape[1]
    attention_mask = None
    if padding_mask is not None:
        attention_mask = self._expand_padding_mask(padding_mask, sequence_length)
    hidden_states, rope_pe = self._apply_positional_encoding(hidden_states)
    for layer in self.layers:
        hidden_states = layer(
            hidden_states=hidden_states,
            attention_mask=attention_mask,
            positional_encoding=rope_pe,
        )
    hidden_states = self.final_normalization(hidden_states)
    return hidden_states