quaterion_models.heads.empty_head module

class EmptyHead(input_embedding_size: int, dropout: float = 0.0)[source]

Bases: EncoderHead

Returns input embeddings without any modification

transform(input_vectors: Tensor) → Tensor[source]

Apply head-specific transformations to the embeddings tensor. Called as part of forward function, but with generic wrappings

Parameters:

input_vectors – Concatenated embeddings of all encoders. Shape: (batch_size, self.input_embedding_size)

Returns:

Final embeddings for a batch – (batch_size, self.output_size)

property output_size: int

training: bool