码迷,mamicode.com
首页 > 其他好文 > 详细

[code] Transformer For Summarization Source Code Reading

时间:2019-07-23 00:17:18      阅读:81      评论:0      收藏:0      [点我收藏+]

标签:debug   asi   ror   正弦   嵌入   框架   distrib   常用   std   

Basic Information

作者:李丕绩(腾讯AI Lab)

模型:Transformer + copy mechanism for abstractive summarization

数据集:CNN/Daily Mail

Parameters

WARNING: IN DEBUGGING MODE
USE COPY MECHANISM
USE COVERAGE MECHANISM
USE AVG NLL as LOSS
USE LEARNABLE W2V EMBEDDING
RNN TYPE: transformer
idx_gpu: 0
norm_clip: 2  # gradient clipping by norm
dim_x: 512
dim_y: 512
len_x: 401
len_y: 101
num_x: 1
num_y: 1
hidden_size: 512
d_ff: 1024
num_heads: 8  # 8头注意力机制
dropout: 0.2
num_layers: 4
label_smoothing: 0.1
alpha: 0.9
beta: 5
batch_size: 5
testing_batch_size: 1
min_len_predict: 35
max_len_predict: 120
max_byte_predict: None
testing_print_size: 500
lr: 0.15
beam_size: 4
max_epoch: 50
print_time: 20  # 每个 epoch 打印信息,以及保存模型的次数
save_epoch: 1
dict_size: 50003  # vocabulary 大小
pad_token_idx: 0
loading train set...
num_files =  13
num_batches =  3

Model Structure

Model(
  (tok_embed): Embedding(50003, 512, padding_idx=0)
  (pos_embed): LearnedPositionalEmbedding(
    (weights): Embedding(1024, 512)
  )
  (enc_layers): ModuleList(
    (0): TransformerLayer(
      (self_attn): MultiheadAttention(
        (out_proj): Linear(in_features=512, out_features=512, bias=True)
      )
      (fc1): Linear(in_features=512, out_features=1024, bias=True)
      (fc2): Linear(in_features=1024, out_features=512, bias=True)
      (attn_layer_norm): LayerNorm()
      (ff_layer_norm): LayerNorm()
    )
    (1): TransformerLayer(
      (self_attn): MultiheadAttention(
        (out_proj): Linear(in_features=512, out_features=512, bias=True)
      )
      (fc1): Linear(in_features=512, out_features=1024, bias=True)
      (fc2): Linear(in_features=1024, out_features=512, bias=True)
      (attn_layer_norm): LayerNorm()
      (ff_layer_norm): LayerNorm()
    )
    (2): TransformerLayer(
      (self_attn): MultiheadAttention(
        (out_proj): Linear(in_features=512, out_features=512, bias=True)
      )
      (fc1): Linear(in_features=512, out_features=1024, bias=True)
      (fc2): Linear(in_features=1024, out_features=512, bias=True)
      (attn_layer_norm): LayerNorm()
      (ff_layer_norm): LayerNorm()
    )
    (3): TransformerLayer(
      (self_attn): MultiheadAttention(
        (out_proj): Linear(in_features=512, out_features=512, bias=True)
      )
      (fc1): Linear(in_features=512, out_features=1024, bias=True)
      (fc2): Linear(in_features=1024, out_features=512, bias=True)
      (attn_layer_norm): LayerNorm()
      (ff_layer_norm): LayerNorm()
    )
  )
  (dec_layers): ModuleList(
    (0): TransformerLayer(
      (self_attn): MultiheadAttention(
        (out_proj): Linear(in_features=512, out_features=512, bias=True)
      )
      (fc1): Linear(in_features=512, out_features=1024, bias=True)
      (fc2): Linear(in_features=1024, out_features=512, bias=True)
      (attn_layer_norm): LayerNorm()
      (ff_layer_norm): LayerNorm()
      (external_attn): MultiheadAttention(
        (out_proj): Linear(in_features=512, out_features=512, bias=True)
      )
      (external_layer_norm): LayerNorm()
    )
    (1): TransformerLayer(
      (self_attn): MultiheadAttention(
        (out_proj): Linear(in_features=512, out_features=512, bias=True)
      )
      (fc1): Linear(in_features=512, out_features=1024, bias=True)
      (fc2): Linear(in_features=1024, out_features=512, bias=True)
      (attn_layer_norm): LayerNorm()
      (ff_layer_norm): LayerNorm()
      (external_attn): MultiheadAttention(
        (out_proj): Linear(in_features=512, out_features=512, bias=True)
      )
      (external_layer_norm): LayerNorm()
    )
    (2): TransformerLayer(
      (self_attn): MultiheadAttention(
        (out_proj): Linear(in_features=512, out_features=512, bias=True)
      )
      (fc1): Linear(in_features=512, out_features=1024, bias=True)
      (fc2): Linear(in_features=1024, out_features=512, bias=True)
      (attn_layer_norm): LayerNorm()
      (ff_layer_norm): LayerNorm()
      (external_attn): MultiheadAttention(
        (out_proj): Linear(in_features=512, out_features=512, bias=True)
      )
      (external_layer_norm): LayerNorm()
    )
    (3): TransformerLayer(
      (self_attn): MultiheadAttention(
        (out_proj): Linear(in_features=512, out_features=512, bias=True)
      )
      (fc1): Linear(in_features=512, out_features=1024, bias=True)
      (fc2): Linear(in_features=1024, out_features=512, bias=True)
      (attn_layer_norm): LayerNorm()
      (ff_layer_norm): LayerNorm()
      (external_attn): MultiheadAttention(
        (out_proj): Linear(in_features=512, out_features=512, bias=True)
      )
      (external_layer_norm): LayerNorm()
    )
  )
  (attn_mask): SelfAttentionMask()
  (emb_layer_norm): LayerNorm()
  (word_prob): WordProbLayer(
    (external_attn): MultiheadAttention(
      (out_proj): Linear(in_features=512, out_features=512, bias=True)
    )
    (proj): Linear(in_features=1536, out_features=50003, bias=True)
  )
  (smoothing): LabelSmoothing()
)

模型结构:

1. 嵌入表示:token embedding,positional embedding
2. encoder:3个blocks(8头注意力)
3. decoder:3个blocks(8头注意力)
4. mask attention层
5. layer normalization:层标准化
6. word probability layer:映射到单词表上的概率分布

Source Code Analysis

1. prepare_data.py

处理数据,得到的结果如下:

拿test set举例,数据集中有11489对 article-summary。数据结构如下:

所有的数据对组成一级列表,每对数据由两个子列表组成,分别代表article 和 summary

article列表和summary中,分为两个子列表,第一个是分词后的序列,第二个是分词之前的原始文本

2. model.py

利用pytorch框架构建了Transformer summarizer模型:

2.1. __initial__

设置模型参数,几个比较重要的:

使用了copy mechanism 和 coverage mechanism;
使用NLL作为loss function;
d_ff size = 1024;
context size = 512;
hidden size = 512;

定义了几个比较常用的结构:

label smoothing
可学习的token embedding,positional embedding
word probability layer
embedding layer normalization 

2.2. structure of encoder & decoder

技术图片

技术图片

可以看到encoder,decoder中都包含多个(4个)基本modules。每个module的结构如下:

技术图片

其中包含了子模块,子模块的下级模块如下:self-attention结构,两个全连接层,attention normalization layer,feed forward nomalization layer:

技术图片

self-attention 模块结构如下图:

技术图片

2.2.1. embedding(transformer.py)

token embedding 用的是 nn.Embedding 在训练的时候进行学习。模块化参数:vocabulary size = 50003,embedding dim = 512

positional embedding 用的还nn.Embedding,随着训练进行学习。模块参数:init_size = 1024(最大的position),embedding dim = 512;参数用正态分布进行随机初始化

作者同样实现了类 SinusoidalPositionalEmbedding 可以将 learnable positional embedding 变为 fixed,即论文中加入位置信息的方式

2.2.2. encoder & decoder

# encoder
self.enc_layers = nn.ModuleList()
for i in range(self.num_layers):
    self.enc_layers.append(TransformerLayer(self.dim_x, self.d_ff, self.num_heads, self.dropout))

# decoder 
self.dec_layers = nn.ModuleList()
for i in range(self.num_layers):
    self.dec_layers.append(TransformerLayer(self.dim_x, self.d_ff, self.num_heads, self.dropout, with_external=True))
        

torch可以用nn.ModuleList()来增加模块的层,此处num_layer = 4 ,即加入4层Transformer stack作为encoder。

定义基本单元的时候TransformerLayer的初始化定义不同。

2.3. encoding

流程:

  1. 获得嵌入表示(token embedding + positional embedding)
  2. 层标准化
  3. dropout
  4. padding mask
  5. 对于N层encoder stack进行编码(利用Transformer单元;Transformer单元之间参数不共享),上一层encoder的输出做为下一层encoder的输入(layer的输入是:序列的嵌入表示x,和padding mask)
  6. 返回最终的编码向量x

2.4. decoding

decoding 此处分为两种情况:

  1. coverage mechanism + copy mechanism
  2. coverage emchanism

流程:

  1. 前者需要比后者多传入两个参数:x_ext, max_ext_len;即拓展此表中的词也录入id了,max_ext_len表示OOV词的个数
  2. 获得嵌入表示(token embedding + positional embedding)
  3. 层标准化
  4. dropout
  5. padding mask
  6. 对于N层decoder stack进行编码,上一层decoder的输出做为下一层decoder的输入(layer的输入是:序列的嵌入表示x,和padding mask,self attention mask,external memories,external padding mask)
  7. 利用final decoder state进行word probability distribution的计算。分为两种计算方式,在WordProb中实现

2.5. word probability projection (WordProbLayer.py)

进行条件判别:

  1. copy mechanism:
    1. 利用external_attention函数(class MultiheadAttention)计算attention。
    2. 输入的是query = decoder final hidden states;key = value = encoder hidden states,返回的是(attention output, attention weights)
    3. 将解码状态,解码输入(的嵌入表示),external_attention的输出进行concatenation;进行线性映射;经过softmax,得到pred
    4. 如果source article中出现了OOV,max_ext_len>0,则需要将pred拼接上一个全零张量,以至于pred的dim3的维度等于fixed vocabulary size + number of OOV
    5. 设置gate,将解码状态,解码输入(的嵌入表示),external_attention的输出进行concatenation;进行线性映射;经过sigmoid,得到gate
    6. 最终的概率分布:pred = gate * pred + (1-gate) * attention_weights
  2. no copy:利用全连接层进行简单的线性映射,再经过softmax激活函数得到单词表上的概率分布
函数参数:scatter_add_(dim, ?indexTensor, ?otherTensor)?→ 输出Tensor
函数用法:selfTensor.scatter_add_(dim, ?indexTensor, ?otherTensor)
# 该函数将?otherTensor?的所有值加到?selfTensor?中,加入位置由?indexTensor?指明

2.6. loss

2.6.1. label_smoothing_loss (label_smoothing.py)

def label_smotthing_loss(self, y_pred, y, y_mask, avg=True):
    seq_len, bsz = y.size()

    y_pred = T.log(y_pred.clamp(min=1e-8))
    loss = self.smoothing(y_pred.view(seq_len * bsz, -1), y.view(seq_len * bsz, -1))
    if avg:
        return loss / T.sum(y_mask)
    else:
        return loss / bsz

loss function的实现中用到了一个clamp夹逼函数:

torch.clamp(input, min, max, out=None) → Tensor

作用是将inpout tensor的数值限制在min到max之间,大于max即为max,小于min即为min,在两者之间不变

然后y_pred(预测的单词对应的概率值)与真实值y送入smoothing函数,作者写了一个类class LabelSmoothing

初始化类的时候需要传入label_smoothing_factor,即padding位在vocabulary中的索引。利用target计算出model prob;

返回model prob与y_pred之间的KL divergence作为loss

2.6.2. negative_log_likelihood

def nll_loss(self, y_pred, y, y_mask, avg=True):
    cost = -T.log(T.gather(y_pred, 2, y.view(y.size(0), y.size(1), 1)))
    cost = cost.view(y.shape)
    y_mask = y_mask.view(y.shape)
    if avg:
        cost = T.sum(cost * y_mask, 0) / T.sum(y_mask, 0)
        else:
            cost = T.sum(cost * y_mask, 0)
            cost = cost.view((y.size(1), -1))
            return T.mean(cost) 

3. transformer.py

3.1. class Transformer

class TransformerLayer(nn.Module):
    
    def __init__(self, embed_dim, ff_embed_dim, num_heads, dropout, with_external=False, weights_dropout = True):
        super(TransformerLayer, self).__init__()
        self.self_attn = MultiheadAttention(embed_dim, num_heads, dropout, weights_dropout)
        self.fc1 = nn.Linear(embed_dim, ff_embed_dim)
        self.fc2 = nn.Linear(ff_embed_dim, embed_dim)
        self.attn_layer_norm = LayerNorm(embed_dim)
        self.ff_layer_norm = LayerNorm(embed_dim)
        self.with_external = with_external
        self.dropout = dropout
        if self.with_external:
            self.external_attn = MultiheadAttention(embed_dim, num_heads, dropout, weights_dropout)
            self.external_layer_norm = LayerNorm(embed_dim)
        self.reset_parameters()
    
    def reset_parameters(self):
        nn.init.normal_(self.fc1.weight, std=0.02)
        nn.init.normal_(self.fc2.weight, std=0.02)
        nn.init.constant_(self.fc1.bias, 0.)
        nn.init.constant_(self.fc2.bias, 0.)

    def forward(self, x, kv = None,
                self_padding_mask = None, self_attn_mask = None,
                external_memories = None, external_padding_mask=None,
                need_weights = False):
        # x: seq_len x bsz x embed_dim
        residual = x  # 残差:add & norm操作中,需要先将residual,以及residual经过feed forward得到的output进行求和,再进行norm的计算
        if kv is None:
            x, self_attn = self.self_attn(query=x, key=x, value=x, key_padding_mask=self_padding_mask, attn_mask=self_attn_mask, need_weights = need_weights)
        else:
            x, self_attn = self.self_attn(query=x, key=kv, value=kv, key_padding_mask=self_padding_mask, attn_mask=self_attn_mask, need_weights = need_weights)

        x = F.dropout(x, p=self.dropout, training=self.training)
        x = self.attn_layer_norm(residual + x)  # 先将residual,以及residual经过feed forward得到的output进行求和,再进行norm的计算

        if self.with_external:
            residual = x
            x, external_attn = self.external_attn(query=x, key=external_memories, value=external_memories, key_padding_mask=external_padding_mask, need_weights = need_weights)
            x = F.dropout(x, p=self.dropout, training=self.training)
            x = self.external_layer_norm(residual + x)
        else:
            external_attn = None

        residual = x
        x = gelu(self.fc1(x))  # 高斯误差线性单元 Gaussian Error Linear Units(GELU)
        x = F.dropout(x, p=self.dropout, training=self.training)
        x = self.fc2(x)  # 全连接层
        x = F.dropout(x, p=self.dropout, training=self.training)
        x = self.ff_layer_norm(residual + x)

        return x, self_attn, external_attn

基本结构:

  1. self attention层(class MultiheadAttention):初始化Transformer单元的时候,需要指定number of head
  2. external attention层(class MultiheadAttention):如果参数with_external为True,计算attention的时候需要考虑外界的输入。即不再是query = key = value了,而key 和 value可能来自source端,这在decoder中需要用到。

  3. 全连接层(两个),形状:(embed_dim, ff_embed_dim),(ff_embed_dim, embed_dim)

  4. attention layer normalization + feedforward normalization(class LayerNorm
  5. dropout
  6. parameter initialization:主要针对的是两个全连接层的weights和bias

功能(forward函数):

  1. 记录下residual。在add & norm操作中,需要先将经过计算得到的output与residual进行求和,再进行normalization
  2. self_attention,用的是class MultiheadAttention,需要提供给forward函数:query,key,value;以及key_padding_mask,atten_mask
  3. dropout
  4. add & attention normalization
  5. 如果参数with_external为True,需要额外进行:
    1. external attention
    2. dropout
    3. add & attention normalization
  6. 再记录residual
  7. 经过全连接层f1
  8. 高斯误差线性单元 Gaussian Error Linear Units(GELU)
  9. dropout
  10. 经过全连接层f2
  11. dropout
  12. add & feedforward normalization

其中激活函数GLUE的定义如下:

def gelu(x):
    cdf = 0.5 * (1.0 + torch.erf(x / math.sqrt(2.0)))
    return cdf*x

3.2. class MultiheadAttention

初始化MultiheadAttention需要几个参数:

  1. attention head count:8
  2. embed_dim (dim_x) : 512
  3. ff_embed_dim:1024

head_dim,即每个注意力头的维度的计算方式为:head_dim = embed_dim // num_heads,前者必须可以为后者所整除

attention 用的还是论文中scaled attention,scaling参数是head_dim开方

对收入、输出的映射:

  1. in_proj_weight:(3*embed_dim, embed_dim)
  2. in_proj_bia:(3*embed_dim),先定义出来,前1/3是Query的,中间1/3是Key的,最后1/3是Value的。后面定义了一个函数_in_proj,根据传入的参数确定需要对qkv中的那几个进行映射,取出来就行了。但是输入映射的参数肯定是从in_proj_weight,in_proj_bia中取的

  3. out_proj:(embed_dim, embed_dim)

对具体情况进行判别,对应地对输入的qkv进行输入映射:

  1. 如果qkv相同,则是self-attention
  2. 如果qkv不同,但是kv相同,则是encoder-decoder attention
  3. 如果qkv均不同,则是一般的attention

对attention weights进行mask,使用的是方法masked_fill_ 输入一个ByteTensor,其中元素为1的位置,对应Tensor中元素会被置0。

对attention weights进行mask,再进行softmax,再进行dropout

attention output是attention weights与value进行bmm (batch matrix multiply),结果再包上一层dropout

进行一次输出映射,得到MultiheadAttention的输出

3.3. class LayerNorm

def forward(self, x):
    u = x.mean(-1, keepdim=True)
    s = (x - u).pow(2).mean(-1, keepdim=True)
    x = (x - u) / torch.sqrt(s + self.eps)
    return self.weight * x + self.bias

流程:

  1. 计算均值

  2. 计算方差

  3. 减去方差,除以标准差

  4. 经过一个线性映射(fully connected layer)

3.4. Positional Embedding

定义了两个类:

  1. 可学习的位置编码:class LearnedPositionalEmbedding
  2. 由正弦函数给出的固定位置编码:class SinusoidalPositionalEmbedding定义与Attention is All You Need 论文中一致

[code] Transformer For Summarization Source Code Reading

标签:debug   asi   ror   正弦   嵌入   框架   distrib   常用   std   

原文地址:https://www.cnblogs.com/lauspectrum/p/11229125.html

(0)
(0)
   
举报
评论 一句话评论(0
登录后才能评论!
© 2014 mamicode.com 版权所有  联系我们:gaon5@hotmail.com
迷上了代码!