Transformer模型架构及PyTorch源码详解（基于Attention is All You Need）

张好好-学习

已于 2022-04-26 11:29:07 修改

阅读量3.1k

点赞数 2

于 2022-04-24 18:59:09 首次发布

本文链接：https://blog.csdn.net/weixin_42756682/article/details/124386522

版权

深度学习专栏收录该内容

12 篇文章 2 订阅

订阅专栏

模型架构

在这里插入图片描述
Encoder：
N个block组成，每个block由一个自注意层和+一个FFN层组成

Decoder：
N个block组成，每个block由一个masked自注意层+交叉注意层+FFN层组成

为什么以第一个自注意层需要对输入的右侧进行mask？ decoder的query的输入是串行的，在测试时前面的query输入时其实是看不见后面的序列，因此应该对其进行mask，以保证当前的判断仅依赖于此前的序列。

交叉注意层——q来自decoder；k，v来自encoder的输出

使用的是点乘注意力得分计算方法：
在这里插入图片描述这里把attention抽象为对 value() 的每个 token进行加权，而加权的weight就是 attentionweight，而 attention weight 就是根据 query和 key 计算得到，其意义为：为了用 value求出 query的结果, 根据 query和 key 来决定注意力应该放在value的哪部分。

为什么要除以√d_k？ 是因为如果d_k太大，点乘的值太大，如果不做scaling，结果就没有加法注意力好。另外，点乘的结果过大，这使得经过softmax之后的梯度很小，不利于反向传播的进行，所以我们通过对点乘的结果进行尺度化。

多头注意力，就是将多个上述部分的结果拼接起来：
在这里插入图片描述

位置编码：
位置编码会随着残差计算向后传播,
本文采用的是sin/cos位置编码，计算公式如下图所示：
在这里插入图片描述

源码

有五个相关类：

Transformer
TransformerEncoder
TransformerDecoder
TransformerEncoderLayer
TransformerDecoderLayer

torch.nn.Transformer

import copy
from typing import Optional, Any, Union, Callable

import torch
from torch import Tensor
from .. import functional as F
from .module import Module
from .activation import MultiheadAttention
from .container import ModuleList
from ..init import xavier_uniform_
from .dropout import Dropout
from .linear import Linear
from .normalization import LayerNorm


class Transformer(Module):
    r"""A transformer model. User is able to modify the attributes as needed. The architecture
    is based on the paper "Attention Is All You Need". Ashish Vaswani, Noam Shazeer,
    Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Lukasz Kaiser, and
    Illia Polosukhin. 2017. Attention is all you need. In Advances in Neural Information
    Processing Systems, pages 6000-6010. Users can build the BERT(https://arxiv.org/abs/1810.04805)
    model with corresponding parameters.

    Args:
        d_model: the number of expected features in the encoder/decoder inputs (default=512).
        nhead: the number of heads in the multiheadattention models (default=8).
        num_encoder_layers: the number of sub-encoder-layers in the encoder (default=6).
        num_decoder_layers: the number of sub-decoder-layers in the decoder (default=6).
        dim_feedforward: the dimension of the feedforward network model (default=2048).
        dropout: the dropout value (default=0.1).
        activation: the activation function of encoder/decoder intermediate layer, can be a string
            ("relu" or "gelu") or a unary callable. Default: relu
        custom_encoder: custom encoder (default=None).
        custom_decoder: custom decoder (default=None).
        layer_norm_eps: the eps value in layer normalization components (default=1e-5).
        batch_first: If ``True``, then the input and output tensors are provided
            as (batch, seq, feature). Default: ``False`` (seq, batch, feature).
        norm_first: if ``True``, encoder and decoder layers will perform LayerNorms before
            other attention and feedforward operations, otherwise after. Default: ``False`` (after).

    Examples::
        >>> transformer_model = nn.Transformer(nhead=16, num_encoder_layers=12)
        >>> src = torch.rand((10, 32, 512))
        >>> tgt = torch.rand((20, 32, 512))
        >>> out = transformer_model(src, tgt)

    Note: A full example to apply nn.Transformer module for the word language model is available in
    https://github.com/pytorch/examples/tree/master/word_language_model
    """
    def __init__(self, d_model: int = 512, nhead: int = 8, num_encoder_layers: int = 6,
                 num_decoder_layers: int = 6, dim_feedforward: int = 2048, dropout: float = 0.1,
                 activation: Union[str, Callable[[Tensor], Tensor]] = F.relu,
                 custom_encoder: Optional[Any] = None, custom_decoder: Optional[Any] = None,
                 layer_norm_eps: float = 1e-5, batch_first: bool = False, norm_first: bool = False,
                 device=None, dtype=None) -> None:
                 
       			pass
       	
	def forward(self, src: Tensor, tgt: Tensor, src_mask: Optional[Tensor] = None, tgt_mask: Optional[Tensor] = None,
                memory_mask: Optional[Tensor] = None, src_key_padding_mask: Optional[Tensor] = None,
                tgt_key_padding_mask: Optional[Tensor] = None, memory_key_padding_mask: Optional[Tensor] = None) -> Tensor:
                
                pass
                
	@staticmethod
    def generate_square_subsequent_mask(sz: int) -> Tensor:
    	pass
    
    def _reset_parameters(self):
    	pass

init

调用及参数

torch.nn.Transformer(d_model=512, nhead=8, num_encoder_layers=6, num_decoder_layers=6, 
					dim_feedforward=2048, dropout=0.1, activation='relu', 
					custom_encoder=None, custom_decoder=None)

参数：
d_model –编码器/解码器输入大小（默认 512）。
nhead –多头注意力模型的头数（默认为8）。
num_encoder_layers –编码器中子编码器层（block）的数量（默认为6）。
num_decoder_layers –解码器中子解码器层（block）的数量（默认为6）。
dim_feedforward –前馈网络模型的中间层维度（默认= 2048）。

dropout –默认值= 0.1。
activation–编码器/解码器中间层的激活函数，relu或gelu（默认值= relu）。
custom_encoder –自定义编码器（默认=None）。
custom_decoder –自定义解码器（默认=None）。

源码

def __init__(self, d_model: int = 512, nhead: int = 8, num_encoder_layers: int = 6,
                 num_decoder_layers: int = 6, dim_feedforward: int = 2048, dropout: float = 0.1,
                 activation: Union[str, Callable[[Tensor], Tensor]] = F.relu,
                 custom_encoder: Optional[Any] = None, custom_decoder: Optional[Any] = None,
                 layer_norm_eps: float = 1e-5, batch_first: bool = False, norm_first: bool = False,
                 device=None, dtype=None) -> None:
        factory_kwargs = {'device': device, 'dtype': dtype}
        super(Transformer, self).__init__()

        if custom_encoder is not None:	//是否自定义编码器
            self.encoder = custom_encoder
        else:
            encoder_layer = TransformerEncoderLayer(d_model, nhead, dim_feedforward, dropout,
                                                    activation, layer_norm_eps, batch_first, norm_first,
                                                    **factory_kwargs)
            encoder_norm = LayerNorm(d_model, eps=layer_norm_eps, **factory_kwargs)
            self.encoder = TransformerEncoder(encoder_layer, num_encoder_layers, encoder_norm)

        if custom_decoder is not None:
            self.decoder = custom_decoder
        else:
            decoder_layer = TransformerDecoderLayer(d_model, nhead, dim_feedforward, dropout,
                                                    activation, layer_norm_eps, batch_first, norm_first,
                                                    **factory_kwargs)
            decoder_norm = LayerNorm(d_model, eps=layer_norm_eps, **factory_kwargs)
            self.decoder = TransformerDecoder(decoder_layer, num_decoder_layers, decoder_norm)

        self._reset_parameters()

        self.d_model = d_model
        self.nhead = nhead

        self.batch_first = batch_first

forward

调用及参数

forward(src, tgt, src_mask=None, tgt_mask=None, memory_mask=None, 
src_key_padding_mask=None, tgt_key_padding_mask=None, memory_key_padding_mask=None)


参数：
src – the sequence to the encoder (required).
tgt – the sequence to the decoder (required).
src_mask – the additive mask for the src sequence (optional).
tgt_mask – the additive mask for the tgt sequence (optional).
memory_mask – the additive mask for the encoder output (optional).
src_key_padding_mask – the ByteTensor mask for src keys per batch (optional).
tgt_key_padding_mask – the ByteTensor mask for tgt keys per batch (optional).
memory_key_padding_mask – the ByteTensor mask for memory keys per batch (optional).

shape:
// S is the source sequence length, T is the target sequence length, N is the batch size, E is the feature number
src: (S, E)(S,E) for unbatched input, (S, N, E)(S,N,E) if batch_first=False or (N, S, E) if batch_first=True.
tgt: (T, E)(T,E) for unbatched input, (T, N, E)(T,N,E) if batch_first=False or (N, T, E) if batch_first=True.
src_mask: (S, S)(S,S).
tgt_mask: (T, T)(T,T).
memory_mask: (T, S)(T,S).
src_key_padding_mask: (S)(S) for unbatched input otherwise (N, S)(N,S).
tgt_key_padding_mask: (T)(T) for unbatched input otherwise (N, T)(N,T).
memory_key_padding_mask: (S)(S) for unbatched input otherwise (N, S)(N,S).

源码

def forward(self, src: Tensor, tgt: Tensor, src_mask: Optional[Tensor] = None, tgt_mask: Optional[Tensor] = None,
                memory_mask: Optional[Tensor] = None, src_key_padding_mask: Optional[Tensor] = None,
                tgt_key_padding_mask: Optional[Tensor] = None, memory_key_padding_mask: Optional[Tensor] = None) -> Tensor:
        r"""Take in and process masked source/target sequences.

        Args:
            src: the sequence to the encoder (required).
            tgt: the sequence to the decoder (required).
            src_mask: the additive mask for the src sequence (optional).
            tgt_mask: the additive mask for the tgt sequence (optional).
            memory_mask: the additive mask for the encoder output (optional).
            src_key_padding_mask: the ByteTensor mask for src keys per batch (optional).
            tgt_key_padding_mask: the ByteTensor mask for tgt keys per batch (optional).
            memory_key_padding_mask: the ByteTensor mask for memory keys per batch (optional).

        Shape:
            - src: :math:`(S, E)` for unbatched input, :math:`(S, N, E)` if `batch_first=False` or
              `(N, S, E)` if `batch_first=True`.
            - tgt: :math:`(T, E)` for unbatched input, :math:`(T, N, E)` if `batch_first=False` or
              `(N, T, E)` if `batch_first=True`.
            - src_mask: :math:`(S, S)`.
            - tgt_mask: :math:`(T, T)`.
            - memory_mask: :math:`(T, S)`.
            - src_key_padding_mask: :math:`(S)` for unbatched input otherwise :math:`(N, S)`.
            - tgt_key_padding_mask: :math:`(T)` for unbatched input otherwise :math:`(N, T)`.
            - memory_key_padding_mask: :math:`(S)` for unbatched input otherwise :math:`(N, S)`.

            Note: [src/tgt/memory]_mask ensures that position i is allowed to attend the unmasked
            positions. If a ByteTensor is provided, the non-zero positions are not allowed to attend
            while the zero positions will be unchanged. If a BoolTensor is provided, positions with ``True``
            are not allowed to attend while ``False`` values will be unchanged. If a FloatTensor
            is provided, it will be added to the attention weight.
            [src/tgt/memory]_key_padding_mask provides specified elements in the key to be ignored by
            the attention. If a ByteTensor is provided, the non-zero positions will be ignored while the zero
            positions will be unchanged. If a BoolTensor is provided, the positions with the
            value of ``True`` will be ignored while the position with the value of ``False`` will be unchanged.

            - output: :math:`(T, E)` for unbatched input, :math:`(T, N, E)` if `batch_first=False` or
              `(N, T, E)` if `batch_first=True`.

            Note: Due to the multi-head attention architecture in the transformer model,
            the output sequence length of a transformer is same as the input sequence
            (i.e. target) length of the decode.

            where S is the source sequence length, T is the target sequence length, N is the
            batch size, E is the feature number

        Examples:
            >>> output = transformer_model(src, tgt, src_mask=src_mask, tgt_mask=tgt_mask)
        """

        is_batched = src.dim() == 3
        if not self.batch_first and src.size(1) != tgt.size(1) and is_batched:
            raise RuntimeError("the batch number of src and tgt must be equal")
        elif self.batch_first and src.size(0) != tgt.size(0) and is_batched:
            raise RuntimeError("the batch number of src and tgt must be equal")

        if src.size(-1) != self.d_model or tgt.size(-1) != self.d_model:
            raise RuntimeError("the feature number of src and tgt must be equal to d_model")

        memory = self.encoder(src, mask=src_mask, src_key_padding_mask=src_key_padding_mask)
        output = self.decoder(tgt, memory, tgt_mask=tgt_mask, memory_mask=memory_mask,
                              tgt_key_padding_mask=tgt_key_padding_mask,
                              memory_key_padding_mask=memory_key_padding_mask)
        return output

torch.nn.TransformerEncoderLayer

class TransformerEncoderLayer(Module):
    r"""TransformerEncoderLayer is made up of self-attn and feedforward network.
    This standard encoder layer is based on the paper "Attention Is All You Need".
    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez,
    Lukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. In Advances in
    Neural Information Processing Systems, pages 6000-6010. Users may modify or implement
    in a different way during application.

    Args:
        d_model: the number of expected features in the input (required).
        nhead: the number of heads in the multiheadattention models (required).
        dim_feedforward: the dimension of the feedforward network model (default=2048).
        dropout: the dropout value (default=0.1).
        activation: the activation function of the intermediate layer, can be a string
            ("relu" or "gelu") or a unary callable. Default: relu
        layer_norm_eps: the eps value in layer normalization components (default=1e-5).
        batch_first: If ``True``, then the input and output tensors are provided
            as (batch, seq, feature). Default: ``False`` (seq, batch, feature).
        norm_first: if ``True``, layer norm is done prior to attention and feedforward
            operations, respectivaly. Otherwise it's done after. Default: ``False`` (after).

    Examples::
        >>> encoder_layer = nn.TransformerEncoderLayer(d_model=512, nhead=8)
        >>> src = torch.rand(10, 32, 512)
        >>> out = encoder_layer(src)

    Alternatively, when ``batch_first`` is ``True``:
        >>> encoder_layer = nn.TransformerEncoderLayer(d_model=512, nhead=8, batch_first=True)
        >>> src = torch.rand(32, 10, 512)
        >>> out = encoder_layer(src)
    """
    __constants__ = ['batch_first', 'norm_first']

    def __init__(self, d_model: int, nhead: int, dim_feedforward: int = 2048, dropout: float = 0.1,
                 activation: Union[str, Callable[[Tensor], Tensor]] = F.relu,
                 layer_norm_eps: float = 1e-5, batch_first: bool = False, norm_first: bool = False,
                 device=None, dtype=None) -> None:


    def __setstate__(self, state):


	def forward(self, src: Tensor, src_mask: Optional[Tensor] = None, src_key_padding_mask: Optional[Tensor] = None) -> Tensor:

    # self-attention block
    def _sa_block(self, x: Tensor,attn_mask: Optional[Tensor], 
    				key_padding_mask: Optional[Tensor]) -> Tensor:

init

调用及参数

torch.nn.TransformerEncoderLayer(d_model, nhead,
			dim_feedforward=2048, dropout=0.1, activation='relu')

参数：
d_model – the number of expected features in the input (required).
nhead – the number of heads in the multiheadattention models (required).
dim_feedforward – the dimension of the feedforward network model (default=2048).
dropout – the dropout value (default=0.1).
activation – the activation function of intermediate layer, relu or gelu (default=relu).

源码

def __init__(self, d_model: int, nhead: int, dim_feedforward: int = 2048, dropout: float = 0.1,
                 activation: Union[str, Callable[[Tensor], Tensor]] = F.relu,
                 layer_norm_eps: float = 1e-5, batch_first: bool = False, norm_first: bool = False,
                 device=None, dtype=None) -> None:
        factory_kwargs = {'device': device, 'dtype': dtype}
        super(TransformerEncoderLayer, self).__init__()
        self.self_attn = MultiheadAttention(d_model, nhead, dropout=dropout, batch_first=batch_first,
                                            **factory_kwargs)
        # Implementation of Feedforward model
        self.linear1 = Linear(d_model, dim_feedforward, **factory_kwargs)
        self.dropout = Dropout(dropout)
        self.linear2 = Linear(dim_feedforward, d_model, **factory_kwargs)

        self.norm_first = norm_first
        self.norm1 = LayerNorm(d_model, eps=layer_norm_eps, **factory_kwargs)
        self.norm2 = LayerNorm(d_model, eps=layer_norm_eps, **factory_kwargs)
        self.dropout1 = Dropout(dropout)
        self.dropout2 = Dropout(dropout)

        # Legacy string support for activation function.
        if isinstance(activation, str):
            self.activation = _get_activation_fn(activation)
        else:
            self.activation = activation

forward

调用及参数

forward(src, src_mask=None, src_key_padding_mask=None)

参数：
src – the sequnce to the encoder layer (required).
src_mask – the mask for the src sequence (optional).
src_key_padding_mask – the mask for the src keys per batch (optional).

源码

def forward(self, src: Tensor, src_mask: Optional[Tensor] = None, src_key_padding_mask: Optional[Tensor] = None) -> Tensor:
        r"""Pass the input through the encoder layer.

        Args:
            src: the sequence to the encoder layer (required).
            src_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.
        """

        # see Fig. 1 of https://arxiv.org/pdf/2002.04745v1.pdf

        x = src
        if self.norm_first:
            x = x + self._sa_block(self.norm1(x), src_mask, src_key_padding_mask)
            x = x + self._ff_block(self.norm2(x))
        else:
            x = self.norm1(x + self._sa_block(x, src_mask, src_key_padding_mask))  //self-attention
            x = self.norm2(x + self._ff_block(x))	//FFN

        return x

	 # self-attention block
    def _sa_block(self, x: Tensor,
                  attn_mask: Optional[Tensor], key_padding_mask: Optional[Tensor]) -> Tensor:
                  //self.self_attn==>MultiheadAttention
        x = self.self_attn(x, x, x,
                           attn_mask=attn_mask,
                           key_padding_mask=key_padding_mask,
                           need_weights=False)[0]
        return self.dropout1(x)

    # feed forward block
    def _ff_block(self, x: Tensor) -> Tensor:
        x = self.linear2(self.dropout(self.activation(self.linear1(x))))
        return self.dropout2(x)

torch.nn.TransformerEncoder

TransformerEncoder is a stack of N encoder layers

class TransformerEncoder(Module):
    r"""TransformerEncoder is a stack of N encoder layers

    Args:
        encoder_layer: an instance of the TransformerEncoderLayer() class (required).
        num_layers: the number of sub-encoder-layers in the encoder (required).
        norm: the layer normalization component (optional).

    Examples::
        >>> encoder_layer = nn.TransformerEncoderLayer(d_model=512, nhead=8)
        >>> transformer_encoder = nn.TransformerEncoder(encoder_layer, num_layers=6)
        >>> src = torch.rand(10, 32, 512)
        >>> out = transformer_encoder(src)
    """
    __constants__ = ['norm']

    def __init__(self, encoder_layer, num_layers, norm=None):


	def forward(self, src: Tensor, mask: Optional[Tensor] = None, src_key_padding_mask: Optional[Tensor] = None) -> Tensor:

init

调用及参数

torch.nn.TransformerEncoder(encoder_layer, num_layers, norm=None)

参数：
coder_layer – TransformerEncoderLayer类的实例（必需）。
num_layers –编码器中的子编码器层数（必填）。
norm –层归一化组件（可选）。

源码

def __init__(self, encoder_layer, num_layers, norm=None):
        super(TransformerEncoder, self).__init__()
        self.layers = _get_clones(encoder_layer, num_layers)
        self.num_layers = num_layers
        self.norm = norm

forward

调用及参数

forward(src, mask=None, src_key_padding_mask=None)

参数：
src – the sequnce 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).

源码

def forward(self, src: Tensor, mask: Optional[Tensor] = None, src_key_padding_mask: Optional[Tensor] = None) -> 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

        for mod in self.layers:
            output = mod(output, src_mask=mask, src_key_padding_mask=src_key_padding_mask)

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

        return output

torch.nn.TransformerDecoderLayer

TransformerDecoderLayer is made up of self-attn, multi-head-attn and feedforward network

class TransformerDecoderLayer(Module):
    r"""TransformerDecoderLayer is made up of self-attn, multi-head-attn and feedforward network.
    This standard decoder layer is based on the paper "Attention Is All You Need".
    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez,
    Lukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. In Advances in
    Neural Information Processing Systems, pages 6000-6010. Users may modify or implement
    in a different way during application.

    Args:
        d_model: the number of expected features in the input (required).
        nhead: the number of heads in the multiheadattention models (required).
        dim_feedforward: the dimension of the feedforward network model (default=2048).
        dropout: the dropout value (default=0.1).
        activation: the activation function of the intermediate layer, can be a string
            ("relu" or "gelu") or a unary callable. Default: relu
        layer_norm_eps: the eps value in layer normalization components (default=1e-5).
        batch_first: If ``True``, then the input and output tensors are provided
            as (batch, seq, feature). Default: ``False`` (seq, batch, feature).
        norm_first: if ``True``, layer norm is done prior to self attention, multihead
            attention and feedforward operations, respectivaly. Otherwise it's done after.
            Default: ``False`` (after).

    Examples::
        >>> decoder_layer = nn.TransformerDecoderLayer(d_model=512, nhead=8)
        >>> memory = torch.rand(10, 32, 512)
        >>> tgt = torch.rand(20, 32, 512)
        >>> out = decoder_layer(tgt, memory)

    Alternatively, when ``batch_first`` is ``True``:
        >>> decoder_layer = nn.TransformerDecoderLayer(d_model=512, nhead=8, batch_first=True)
        >>> memory = torch.rand(32, 10, 512)
        >>> tgt = torch.rand(32, 20, 512)
        >>> out = decoder_layer(tgt, memory)
    """
    __constants__ = ['batch_first', 'norm_first']

    def __init__(self, d_model: int, nhead: int, dim_feedforward: int = 2048, dropout: float = 0.1,
                 activation: Union[str, Callable[[Tensor], Tensor]] = F.relu,
                 layer_norm_eps: float = 1e-5, batch_first: bool = False, norm_first: bool = False,
                 device=None, dtype=None) -> None:


    def __setstate__(self, state):

	def forward(self, tgt: Tensor, memory: Tensor, tgt_mask: Optional[Tensor] = None, memory_mask: Optional[Tensor] = None,
                tgt_key_padding_mask: Optional[Tensor] = None, memory_key_padding_mask: Optional[Tensor] = None) -> Tensor:



    # self-attention block
    def _sa_block(self, x: Tensor,
                  attn_mask: Optional[Tensor], key_padding_mask: Optional[Tensor]) -> Tensor:

    # multihead attention block
    def _mha_block(self, x: Tensor, mem: Tensor,
                   attn_mask: Optional[Tensor], key_padding_mask: Optional[Tensor]) -> Tensor:

    # feed forward block
    def _ff_block(self, x: Tensor) -> Tensor:

init

调用及参数

torch.nn.TransformerDecoderLayer(d_model, nhead, dim_feedforward=2048, dropout=0.1, activation='relu')

参数：
d_model – the number of expected features in the input (required).
nhead – the number of heads in the multiheadattention models (required).
dim_feedforward – the dimension of the feedforward network model (default=2048).
dropout – the dropout value (default=0.1).
activation – the activation function of intermediate layer, relu or gelu (default=relu).

源码

def __init__(self, d_model: int, nhead: int, dim_feedforward: int = 2048, dropout: float = 0.1,
                 activation: Union[str, Callable[[Tensor], Tensor]] = F.relu,
                 layer_norm_eps: float = 1e-5, batch_first: bool = False, norm_first: bool = False,
                 device=None, dtype=None) -> None:
        factory_kwargs = {'device': device, 'dtype': dtype}
        super(TransformerDecoderLayer, self).__init__()
        
        ## 在Decoder中有两个MultiheadAttention，一个是masked self-Attention；一个是cross Attention
        self.self_attn = MultiheadAttention(d_model, nhead, dropout=dropout, batch_first=batch_first,
                                            **factory_kwargs)
        self.multihead_attn = MultiheadAttention(d_model, nhead, dropout=dropout, batch_first=batch_first,
                                                 **factory_kwargs)
        # Implementation of Feedforward model
        self.linear1 = Linear(d_model, dim_feedforward, **factory_kwargs)
        self.dropout = Dropout(dropout)
        self.linear2 = Linear(dim_feedforward, d_model, **factory_kwargs)

        self.norm_first = norm_first
        ## 因为有三个模块，所以此处定义三对，norm、dropout
        self.norm1 = LayerNorm(d_model, eps=layer_norm_eps, **factory_kwargs)
        self.norm2 = LayerNorm(d_model, eps=layer_norm_eps, **factory_kwargs)
        self.norm3 = LayerNorm(d_model, eps=layer_norm_eps, **factory_kwargs)
        self.dropout1 = Dropout(dropout)
        self.dropout2 = Dropout(dropout)
        self.dropout3 = Dropout(dropout)

        # Legacy string support for activation function.
        if isinstance(activation, str):
            self.activation = _get_activation_fn(activation)
        else:
            self.activation = activation

forward

调用及参数

forward(tgt, memory, tgt_mask=None, memory_mask=None, 
tgt_key_padding_mask=None, memory_key_padding_mask=None)

参数：
tgt – the sequence to the decoder layer (required).
memory – the sequnce from the last layer of the encoder (required).
tgt_mask – the mask for the tgt sequence (optional).
memory_mask – the mask for the memory sequence (optional).
tgt_key_padding_mask – the mask for the tgt keys per batch (optional).
memory_key_padding_mask – the mask for the memory keys per batch (optional).

源码

def forward(self, tgt: Tensor, memory: Tensor, tgt_mask: Optional[Tensor] = None, memory_mask: Optional[Tensor] = None,
                tgt_key_padding_mask: Optional[Tensor] = None, memory_key_padding_mask: Optional[Tensor] = None) -> Tensor:
        r"""Pass the inputs (and mask) through the decoder layer.

        Args:
            tgt: the sequence to the decoder layer (required).
            memory: the sequence from the last layer of the encoder (required).
            tgt_mask: the mask for the tgt sequence (optional).
            memory_mask: the mask for the memory sequence (optional).
            tgt_key_padding_mask: the mask for the tgt keys per batch (optional).
            memory_key_padding_mask: the mask for the memory keys per batch (optional).

        Shape:
            see the docs in Transformer class.
        """
        # see Fig. 1 of https://arxiv.org/pdf/2002.04745v1.pdf

        x = tgt
        if self.norm_first:
            x = x + self._sa_block(self.norm1(x), tgt_mask, tgt_key_padding_mask)
            x = x + self._mha_block(self.norm2(x), memory, memory_mask, memory_key_padding_mask)
            x = x + self._ff_block(self.norm3(x))
        else:	## 默认是执行这里
            x = self.norm1(x + self._sa_block(x, tgt_mask, tgt_key_padding_mask))
            x = self.norm2(x + self._mha_block(x, memory, memory_mask, memory_key_padding_mask))
            x = self.norm3(x + self._ff_block(x))

        return x


    # self-attention block
    def _sa_block(self, x: Tensor,
                  attn_mask: Optional[Tensor], key_padding_mask: Optional[Tensor]) -> Tensor:
        x = self.self_attn(x, x, x,				## 自注意层，因此query, key, value都是x
                           attn_mask=attn_mask,
                           key_padding_mask=key_padding_mask,
                           need_weights=False)[0]
        return self.dropout1(x)

    # multihead attention block
    def _mha_block(self, x: Tensor, mem: Tensor,
                   attn_mask: Optional[Tensor], key_padding_mask: Optional[Tensor]) -> Tensor:
        x = self.multihead_attn(x, mem, mem,		## 交叉注意层，query是x； key, value是encoder的输出
                                attn_mask=attn_mask,
                                key_padding_mask=key_padding_mask,
                                need_weights=False)[0]
        return self.dropout2(x)

    # feed forward block
    def _ff_block(self, x: Tensor) -> Tensor:
        x = self.linear2(self.dropout(self.activation(self.linear1(x))))
        return self.dropout3(x)

torch.nn.TransformerDecoder

transformerDecoder是N个TransformerDecoderLayer的堆叠

class TransformerDecoder(Module):
    r"""TransformerDecoder is a stack of N decoder layers

    Args:
        decoder_layer: an instance of the TransformerDecoderLayer() class (required).
        num_layers: the number of sub-decoder-layers in the decoder (required).
        norm: the layer normalization component (optional).

    Examples::
        >>> decoder_layer = nn.TransformerDecoderLayer(d_model=512, nhead=8)
        >>> transformer_decoder = nn.TransformerDecoder(decoder_layer, num_layers=6)
        >>> memory = torch.rand(10, 32, 512)
        >>> tgt = torch.rand(20, 32, 512)
        >>> out = transformer_decoder(tgt, memory)
    """
    __constants__ = ['norm']

    def __init__(self, decoder_layer, num_layers, norm=None):

	def forward(self, tgt: Tensor, memory: Tensor, tgt_mask: Optional[Tensor] = None,
                memory_mask: Optional[Tensor] = None, tgt_key_padding_mask: Optional[Tensor] = None,
                memory_key_padding_mask: Optional[Tensor] = None) -> Tensor:

init

调用及参数

torch.nn.TransformerDecoder(decoder_layer, num_layers, norm=None)

参数：
decoder_layer – TransformerDecoderLayer（）类的实例（必需）。
num_layers –解码器中子解码器层的数量（必需）。
norm –层归一化组件（可选）。

源码

def __init__(self, decoder_layer, num_layers, norm=None):
	super(TransformerDecoder, self).__init__()
    self.layers = _get_clones(decoder_layer, num_layers)
    self.num_layers = num_layers
    self.norm = norm

forward

调用及参数

forward(tgt, memory, tgt_mask=None, memory_mask=None, 
		tgt_key_padding_mask=None, memory_key_padding_mask=None)

参数：
tgt – the sequence to the decoder (required).
memory – the sequnce from the last layer of the encoder (required).
tgt_mask – the mask for the tgt sequence (optional).
memory_mask – the mask for the memory sequence (optional).
tgt_key_padding_mask – the mask for the tgt keys per batch (optional).
memory_key_padding_mask – the mask for the memory keys per batch (optional).

源码

def forward(self, tgt: Tensor, memory: Tensor, tgt_mask: Optional[Tensor] = None,
                memory_mask: Optional[Tensor] = None, tgt_key_padding_mask: Optional[Tensor] = None,
                memory_key_padding_mask: Optional[Tensor] = None) -> Tensor:
        r"""Pass the inputs (and mask) through the decoder layer in turn.

        Args:
            tgt: the sequence to the decoder (required).
            memory: the sequence from the last layer of the encoder (required).
            tgt_mask: the mask for the tgt sequence (optional).
            memory_mask: the mask for the memory sequence (optional).
            tgt_key_padding_mask: the mask for the tgt keys per batch (optional).
            memory_key_padding_mask: the mask for the memory keys per batch (optional).

        Shape:
            see the docs in Transformer class.
        """
        output = tgt

        for mod in self.layers:
            output = mod(output, memory, tgt_mask=tgt_mask,
                         memory_mask=memory_mask,
                         tgt_key_padding_mask=tgt_key_padding_mask,
                         memory_key_padding_mask=memory_key_padding_mask)

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

        return output

其它相关函数

def _get_clones(module, N):
    return ModuleList([copy.deepcopy(module) for i in range(N)])


def _get_activation_fn(activation):
    if activation == "relu":
        return F.relu
    elif activation == "gelu":
        return F.gelu

    raise RuntimeError("activation should be relu/gelu, not {}".format(activation))

Attention部分讲解

注意力函数可以描述为：将query和一组键值（key，value）对映射为输出output，其中query、key、value和output都是向量。output由value的加权和计算得到，其中分配给每个value的权重由query与相应key的兼容函数计算得到。

在本文中计算权重的方法就是 “Scaled Dot-Product Attention"
在这里插入图片描述

简单实现：

在Encoder中的自注意层：
K,Q,V都是input embedding+pos embedding经过三个映射得到的。

在Decoder中：

第一个自注意层：
K,Q,V都是input embedding+pos embedding经过三个映射得到的。
第二个交叉注意层：
Q是Decoder前面的输出经过映射得到的；K，V分别是是encoder的输出经过两种映射得到的。

参考链接

张好好-学习

关注

2
点赞
踩
18

收藏

觉得还不错? 一键收藏
0
评论
Transformer模型架构及PyTorch源码详解（基于Attention is All You Need）

模型架构Encoder：N个block组成，每个block由一个自注意层和+一个FFN层组成Decoder：N个block组成，每个block由一个masked自注意层+交叉注意层+FFN层组成交叉注意层——q来自decoder；k，v来自encoder的输出位置编码：位置编码会随着残差计算向后传播源码有五个相关类：TransformerTransformerEncoderTransformerDecoderTransformerEncoderLayerTransformer
复制链接

扫一扫