core.transformer
================

.. py:module:: core.transformer

.. autoapi-nested-parse::

   Encoder-only transformer model for neural user simulator.



Classes
-------

.. autoapisummary::

   core.transformer.PositionalEncoding
   core.transformer.TransformerEncoderModel


Module Contents
---------------

.. py:class:: PositionalEncoding(d_model: int, dropout: float = 0.1, max_len: int = 5000, **kwargs)

   Bases: :py:obj:`torch.nn.Module`


   Initializes positional encoding layer.

   :param d_model: Dimension of the model.
   :param dropout: Dropout rate. Defaults to 0.1.
   :param max_len: Maximum length of the input sequence. Defaults to 5000.


   .. py:method:: forward(x: torch.Tensor) -> torch.Tensor

      Performs forward pass.

      :param x: Input tensor.

      :returns: Positional encoded tensor.



.. py:class:: TransformerEncoderModel(input_dim: int, output_dim: int, nhead: int, hidden_dim: int, num_encoder_layers: int, num_token: int, dropout: float = 0.5)

   Bases: :py:obj:`torch.nn.Module`


   Initializes a encoder-only transformer model.

   :param input_dim: Size of the input vector.
   :param output_dim: Size of the output vector.
   :param nhead: Number of heads.
   :param hidden_dim: Hidden dimension.
   :param num_encoder_layers: Number of encoder layers.
   :param num_token: Number of tokens in the vocabulary.
   :param dropout: Dropout rate. Defaults to 0.5.


   .. py:method:: init_weights() -> None

      Initializes weights of the network.



   .. py:method:: forward(src: torch.Tensor, src_mask: torch.Tensor = None) -> torch.Tensor

      Performs forward pass.

      :param src: Source tensor.
      :param src_mask: Mask tensor.

      :returns: Output tensor.