10.6. The Encoder-Decoder Architecture¶ Open the notebook in SageMaker Studio Lab
In general seq2seq problems like machine translation (Section 10.5), inputs and outputs are of varying lengths that are unaligned. The standard approach to handling this sort of data is to design an encoder-decoder architecture (Fig. 10.6.1) consisting of two major components: an encoder that takes a variable-length sequence as input, and a decoder that acts as a conditional language model, taking in the encoded input and the leftwards context of the target sequence and predicting the subsequent token in the target sequence.
Fig. 10.6.1 The encoder-decoder architecture.¶
Let’s take machine translation from English to French as an example. Given an input sequence in English: “They”, “are”, “watching”, “.”, this encoder-decoder architecture first encodes the variable-length input into a state, then decodes the state to generate the translated sequence, token by token, as output: “Ils”, “regardent”, “.”. Since the encoder-decoder architecture forms the basis of different seq2seq models in subsequent sections, this section will convert this architecture into an interface that will be implemented later.
from torch import nn
from d2l import torch as d2l
from mxnet.gluon import nn
from d2l import mxnet as d2l
from flax import linen as nn
from d2l import jax as d2l
No GPU/TPU found, falling back to CPU. (Set TF_CPP_MIN_LOG_LEVEL=0 and rerun for more info.)
import tensorflow as tf
from d2l import tensorflow as d2l
10.6.1. Encoder¶
In the encoder interface, we just specify that the encoder takes
variable-length sequences as input X. The implementation will be
provided by any model that inherits this base Encoder class.
class Encoder(nn.Module): #@save
"""The base encoder interface for the encoder-decoder architecture."""
def __init__(self):
super().__init__()
# Later there can be additional arguments (e.g., length excluding padding)
def forward(self, X, *args):
raise NotImplementedError
class Encoder(nn.Block): #@save
"""The base encoder interface for the encoder-decoder architecture."""
def __init__(self):
super().__init__()
# Later there can be additional arguments (e.g., length excluding padding)
def forward(self, X, *args):
raise NotImplementedError
class Encoder(nn.Module): #@save
"""The base encoder interface for the encoder-decoder architecture."""
def setup(self):
raise NotImplementedError
# Later there can be additional arguments (e.g., length excluding padding)
def __call__(self, X, *args):
raise NotImplementedError
class Encoder(tf.keras.layers.Layer): #@save
"""The base encoder interface for the encoder-decoder architecture."""
def __init__(self):
super().__init__()
# Later there can be additional arguments (e.g., length excluding padding)
def call(self, X, *args):
raise NotImplementedError
10.6.2. Decoder¶
In the following decoder interface, we add an additional init_state
method to convert the encoder output (enc_all_outputs) into the
encoded state. Note that this step may require extra inputs, such as the
valid length of the input, which was explained in
Section 10.5. To generate a variable-length
sequence token by token, every time the decoder may map an input (e.g.,
the generated token at the previous time step) and the encoded state
into an output token at the current time step.
class Decoder(nn.Module): #@save
"""The base decoder interface for the encoder-decoder architecture."""
def __init__(self):
super().__init__()
# Later there can be additional arguments (e.g., length excluding padding)
def init_state(self, enc_all_outputs, *args):
raise NotImplementedError
def forward(self, X, state):
raise NotImplementedError
class Decoder(nn.Block): #@save
"""The base decoder interface for the encoder-decoder architecture."""
def __init__(self):
super().__init__()
# Later there can be additional arguments (e.g., length excluding padding)
def init_state(self, enc_all_outputs, *args):
raise NotImplementedError
def forward(self, X, state):
raise NotImplementedError
class Decoder(nn.Module): #@save
"""The base decoder interface for the encoder-decoder architecture."""
def setup(self):
raise NotImplementedError
# Later there can be additional arguments (e.g., length excluding padding)
def init_state(self, enc_all_outputs, *args):
raise NotImplementedError
def __call__(self, X, state):
raise NotImplementedError
class Decoder(tf.keras.layers.Layer): #@save
"""The base decoder interface for the encoder-decoder architecture."""
def __init__(self):
super().__init__()
# Later there can be additional arguments (e.g., length excluding padding)
def init_state(self, enc_all_outputs, *args):
raise NotImplementedError
def call(self, X, state):
raise NotImplementedError
10.6.3. Putting the Encoder and Decoder Together¶
In the forward propagation, the output of the encoder is used to produce the encoded state, and this state will be further used by the decoder as one of its input.
class EncoderDecoder(d2l.Classifier): #@save
"""The base class for the encoder-decoder architecture."""
def __init__(self, encoder, decoder):
super().__init__()
self.encoder = encoder
self.decoder = decoder
def forward(self, enc_X, dec_X, *args):
enc_all_outputs = self.encoder(enc_X, *args)
dec_state = self.decoder.init_state(enc_all_outputs, *args)
# Return decoder output only
return self.decoder(dec_X, dec_state)[0]
class EncoderDecoder(d2l.Classifier): #@save
"""The base class for the encoder-decoder architecture."""
def __init__(self, encoder, decoder):
super().__init__()
self.encoder = encoder
self.decoder = decoder
def forward(self, enc_X, dec_X, *args):
enc_all_outputs = self.encoder(enc_X, *args)
dec_state = self.decoder.init_state(enc_all_outputs, *args)
# Return decoder output only
return self.decoder(dec_X, dec_state)[0]
class EncoderDecoder(d2l.Classifier): #@save
"""The base class for the encoder-decoder architecture."""
encoder: nn.Module
decoder: nn.Module
training: bool
def __call__(self, enc_X, dec_X, *args):
enc_all_outputs = self.encoder(enc_X, *args, training=self.training)
dec_state = self.decoder.init_state(enc_all_outputs, *args)
# Return decoder output only
return self.decoder(dec_X, dec_state, training=self.training)[0]
class EncoderDecoder(d2l.Classifier): #@save
"""The base class for the encoder-decoder architecture."""
def __init__(self, encoder, decoder):
super().__init__()
self.encoder = encoder
self.decoder = decoder
def call(self, enc_X, dec_X, *args):
enc_all_outputs = self.encoder(enc_X, *args, training=True)
dec_state = self.decoder.init_state(enc_all_outputs, *args)
# Return decoder output only
return self.decoder(dec_X, dec_state, training=True)[0]
In the next section, we will see how to apply RNNs to design seq2seq models based on this encoder-decoder architecture.
10.6.4. Summary¶
Encoder-decoder architectures can handle inputs and outputs that both consist of variable-length sequences and thus are suitable for seq2seq problems such as machine translation. The encoder takes a variable-length sequence as input and transforms it into a state with a fixed shape. The decoder maps the encoded state of a fixed shape to a variable-length sequence.
10.6.5. Exercises¶
Suppose that we use neural networks to implement the encoder-decoder architecture. Do the encoder and the decoder have to be the same type of neural network?
Besides machine translation, can you think of another application where the encoder-decoder architecture can be applied?