Advanced Deep Learning with Python

Putting it all together

We'll continue with the main EncoderDecoder class:

class EncoderDecoder(torch.nn.Module):    def __init__(self,                 encoder: Encoder,                 decoder: Decoder,                 source_embeddings: torch.nn.Sequential,                 target_embeddings: torch.nn.Sequential):        super(EncoderDecoder, self).__init__()        self.encoder = encoder        self.decoder = decoder        self.source_embeddings = source_embeddings        self.target_embeddings = target_embeddings    def forward(self, source, target, source_mask, target_mask):        encoder_output = self.encoder(            x=self.source_embeddings(source),            mask=source_mask)        return self.decoder(            x=self.target_embeddings(target),            encoder_states=encoder_output,            source_mask=source_mask,            target_mask=target_mask)

It combines the Encoder, Decoder, and source_embeddings/target_embeddings ...