Method forward

pointersect/models/network_transformer.py:199–227

r"""Pass the input through the encoder layers in turn. Args: src: the sequence to the encoder (required). mask: the mask for the src sequence (optional). src_key_padding_mask: the mask for the src keys per batch (optional). Shape: see

(
            self,
            src: Tensor,
            mask: Optional[Tensor] = None,
            src_key_padding_mask: Optional[Tensor] = None,
    )

        self.norm = norm

    def forward(
            self,
            src: Tensor,
            mask: Optional[Tensor] = None,
            src_key_padding_mask: Optional[Tensor] = None,
    ) -> T.Union[Tensor, Tensor]:
        r"""Pass the input through the encoder layers in turn.

        Args:
            src: the sequence to the encoder (required).
            mask: the mask for the src sequence (optional).
            src_key_padding_mask: the mask for the src keys per batch (optional).

        Shape:
            see the docs in Transformer class.
        """
        output = src
        output_weights_list = []
        for mod in self.layers:
            output, output_weights = mod(output, src_mask=mask, src_key_padding_mask=src_key_padding_mask)
            output_weights_list.append(output_weights)  # ()

        if self.norm is not None:
            output = self.norm(output)

        # Rick: potential improvement, not used during training, saved as list of tensor
        output_weights = torch.stack(output_weights_list, dim=-1)

        return output, output_weights


def _get_clones(module, N):