Transformer[vanilla]
目录
0 引言
transformer这个框架现在可谓是遍地开花，继最开始的AE，CNN，RNN，到现在的transformer，该框架从nlp席卷CV，乃至ASR领域。
本文以The Illustrated Transformer【译】和The Annotated Transformer为来源，主要从总到分的角度去阅读代码。
其实就是觉得The Annotated Transformer写的非常好，但是诸多教程都喜欢先展示一堆材料，然后最后告诉你组装结果；感觉不是很符合自己的理解习惯，因为到了最后组装的时候才发现，前面一堆碎片
1 基于最外层进行示意&一个简单的demo
将The Annotated Transformer里面的demo外部先展现，该demo训练的是一个copy任务，即训练好后，给模型输入一个句子（就是一行数字，表示词索引id），期待输出相同的句子，可以看成是机器翻译任务。
1.1 包引用
import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F
import math, copy, time
from torch.autograd import Variable
import matplotlib.pyplot as plt
import seaborn
seaborn.set_context(context="talk")
这个没得说。
1.2 构建模型 make_model
大多数神经网络翻译模型，都是encoder-decoder 结构，如最经典的seq2seq，将原始向量映射到中间特征向量，然后通过中间特征向量解码成目标列，所以这里核心就是EncoderDecoder部分，以它为拼图的中心
def make_model(src_vocab, tgt_vocab, N=6,
d_model=512, d_ff=2048, h=8, dropout=0.1):
"Helper: Construct a model from hyperparameters."
# 1-方便模块复制
c = copy.deepcopy
# 2-三个额外的组件：多头注意力；逐位置前向；位置编码
attn = MultiHeadedAttention(h, d_model)
ff = PositionwiseFeedForward(d_model, d_ff, dropout)
position = PositionalEncoding(d_model, dropout)
# 3-模型结构
model = EncoderDecoder(
Encoder(EncoderLayer(d_model, c(attn), c(ff), dropout), N),
Decoder(DecoderLayer(d_model, c(attn), c(attn),
c(ff), dropout), N),
nn.Sequential(Embeddings(d_model, src_vocab), c(position)),
nn.Sequential(Embeddings(d_model, tgt_vocab), c(position)),
Generator(d_model, tgt_vocab))
# This was important from their code.
# Initialize parameters with Glorot / fan_avg.
# 4-模型中可训练参数初始化
for p in model.parameters():
if p.dim() > 1:
nn.init.xavier_uniform(p)
return model
上述代码可分成2部分：三个额外的组件：
- 多头注意力；逐位置前向；位置编码；
- 模型结构：编码解码器，其中分为编码器、解码器、序列层1，序列层2，生成器
1.3 构造数据 data_gen
主要是造些假数据，
def data_gen(V, batch, nbatches):
# V是Vocab，单词本的大小
"Generate random data for a src-tgt copy task."
for i in range(nbatches):
data = torch.from_numpy(np.random.randint(1, V, size=(batch, 10)))
data[:, 0] = 1
src = Variable(data, requires_grad=False)
tgt = Variable(data, requires_grad=False)
yield Batch(src, tgt, 0)
1.4 训练
1.4.1 简单的loss计算 SimpleLossCompute
class SimpleLossCompute:
"A simple loss compute and train function."
def __init__(self, generator, criterion, opt=None):
self.generator = generator
self.criterion = criterion
self.opt = opt
def __call__(self, x, y, norm):
# 获取x对应的概率值
x = self.generator(x)
# 计算loss
loss = self.criterion(x.contiguous().view(-1, x.size(-1)),
y.contiguous().view(-1)) / norm
# 反向bp计算，获取梯度值
loss.backward()
if self.opt is not None:
self.opt.step() # 梯度更新到模型
self.opt.optimizer.zero_grad() # 清零梯度值，准备用于下一次梯度计算
return loss.data[0] * norm
1.4.2 训练过程
# Train the simple copy task.
V = 11
# 这里增加了标签平滑的函数
criterion = LabelSmoothing(size=V, padding_idx=0, smoothing=0.0)
model = make_model(V, V, N=2)
#增加了优化器的封装
model_opt = NoamOpt(model.src_embed[0].d_model, 1, 400,
torch.optim.Adam(model.parameters(), lr=0, betas=(0.9, 0.98), eps=1e-9))
for epoch in range(10):
model.train()
run_epoch(data_gen(V, 30, 20), model,
# 这里增加了简单loss计算函数
SimpleLossCompute(model.generator, criterion, model_opt))
model.eval()
print(run_epoch(data_gen(V, 30, 5), model,
SimpleLossCompute(model.generator, criterion, None)))
输出如下
Epoch Step: 1 Loss: 3.023465 Tokens per Sec: 403.074173
Epoch Step: 1 Loss: 1.920030 Tokens per Sec: 641.689380
1.9274832487106324
Epoch Step: 1 Loss: 1.940011 Tokens per Sec: 432.003378
Epoch Step: 1 Loss: 1.699767 Tokens per Sec: 641.979665
1.657595729827881
Epoch Step: 1 Loss: 1.860276 Tokens per Sec: 433.320240
Epoch Step: 1 Loss: 1.546011 Tokens per Sec: 640.537198
1.4888023376464843
Epoch Step: 1 Loss: 1.682198 Tokens per Sec: 432.092305
Epoch Step: 1 Loss: 1.313169 Tokens per Sec: 639.441857
1.3485562801361084
Epoch Step: 1 Loss: 1.278768 Tokens per Sec: 433.568756
Epoch Step: 1 Loss: 1.062384 Tokens per Sec: 642.542067
0.9853351473808288
Epoch Step: 1 Loss: 1.269471 Tokens per Sec: 433.388727
Epoch Step: 1 Loss: 0.590709 Tokens per Sec: 642.862135
0.5686767101287842
Epoch Step: 1 Loss: 0.997076 Tokens per Sec: 433.009746
Epoch Step: 1 Loss: 0.343118 Tokens per Sec: 642.288427
0.34273059368133546
Epoch Step: 1 Loss: 0.459483 Tokens per Sec: 434.594030
Epoch Step: 1 Loss: 0.290385 Tokens per Sec: 642.519464
0.2612409472465515
Epoch Step: 1 Loss: 1.031042 Tokens per Sec: 434.557008
Epoch Step: 1 Loss: 0.437069 Tokens per Sec: 643.630322
0.4323212027549744
Epoch Step: 1 Loss: 0.617165 Tokens per Sec: 436.652626
Epoch Step: 1 Loss: 0.258793 Tokens per Sec: 644.372296
0.27331129014492034
1.5 推理 greedy_decode
def greedy_decode(model, src, src_mask, max_len, start_symbol):
# 调用模型的编码器将输入的 一句话进行编码
memory = model.encode(src, src_mask)
ys = torch.ones(1, 1).fill_(start_symbol).type_as(src.data)
for i in range(max_len-1):
# 不断地进行解码生成下一个词直到遇到用户设定的最大限制
out = model.decode(memory, src_mask,
Variable(ys),
Variable(subsequent_mask(ys.size(1))
.type_as(src.data)))
# 调用模型的生成器获取对应的概率
prob = model.generator(out[:, -1])
_, next_word = torch.max(prob, dim = 1)
next_word = next_word.data[0]
ys = torch.cat([ys,
torch.ones(1, 1).type_as(src.data).fill_(next_word)], dim=1)
return ys
model.eval()
# src是输入
src = Variable(torch.LongTensor([[1,2,3,4,5,6,7,8,9,10]]) )
src_mask = Variable(torch.ones(1, 1, 10) )
# 通过模型推理打印输出
print(greedy_decode(model, src, src_mask, max_len=10, start_symbol=1))
输出如下：
2 EncoderDecoder及内部部分
2.1 EncoderDecoder结构
class EncoderDecoder(nn.Module):
"""
A standard Encoder-Decoder architecture. Base for this and many
other models.
"""
#接受1）编码器、2）解码器、3）源embedding、4）目标embedding、5）生成器
def __init__(self, encoder, decoder, src_embed, tgt_embed, generator):
super(EncoderDecoder, self).__init__()
self.encoder = encoder
self.decoder = decoder
self.src_embed = src_embed
self.tgt_embed = tgt_embed
self.generator = generator
#1）先对源及源mask进行编码；2）再对结果进行解码：【result，src_mask; tgt,tgt_mask】
def forward(self, src, tgt, src_mask, tgt_mask):
"Take in and process masked src and target sequences."
return self.decode(self.encode(src, src_mask), src_mask,
tgt, tgt_mask)
#编码器调用
def encode(self, src, src_mask):
return self.encoder(self.src_embed(src), src_mask)
#解码器调用
def decode(self, memory, src_mask, tgt, tgt_mask):
return self.decoder(self.tgt_embed(tgt), memory, src_mask, tgt_mask)
2.2 前置函数及层
2.2.1 复制模块函数 clones
clones就是便携式的基于list增加多个相同的模块组件
def clones(module, N):
"Produce N identical layers."
return nn.ModuleList([copy.deepcopy(module) for _ in range(N)])
2.2.2 层归一化 LayerNorm
LN的公式：$$a*\frac{x-\bar{x}}{var(x)+eps}+b$$
class LayerNorm(nn.Module):
"Construct a layernorm module (See citation for details)."
def __init__(self, features, eps=1e-6):
super(LayerNorm, self).__init__()
self.a_2 = nn.Parameter(torch.ones(features))
self.b_2 = nn.Parameter(torch.zeros(features))
self.eps = eps
def forward(self, x):
mean = x.mean(-1, keepdim=True)
std = x.std(-1, keepdim=True)
return self.a_2 * (x - mean) / (std + self.eps) + self.b_2
2.2.3 子层连接 SublayerConnection
class SublayerConnection(nn.Module):
"""
A residual connection followed by a layer norm.
Note for code simplicity the norm is first as opposed to last.
"""
def __init__(self, size, dropout):
super(SublayerConnection, self).__init__()
self.norm = LayerNorm(size)
self.dropout = nn.Dropout(dropout)
def forward(self, x, sublayer):
"Apply residual connection to any sublayer with the same size."
return x + self.dropout(sublayer(self.norm(x)))
2.2.4 嵌入 Embeddings
将输入单词和输出单词都转换成embedding，维度为\(d_{model}\),并对这些embedding的权重放大\(\sqrt{d_{model}}\)倍
class Embeddings(nn.Module):
def __init__(self, d_model, vocab):
super(Embeddings, self).__init__()
self.lut = nn.Embedding(vocab, d_model)
self.d_model = d_model
def forward(self, x):
return self.lut(x) * math.sqrt(self.d_model)
2.2.5 subsequent_mask组件及其例子
构建mask，保证只要下三角部分
def subsequent_mask(size):
"Mask out subsequent positions."
attn_shape = (1, size, size)
subsequent_mask = np.triu(np.ones(attn_shape), k=1).astype('uint8')
return torch.from_numpy(subsequent_mask) == 0
plt.figure(figsize=(5,5))
plt.imshow(subsequent_mask(20)[0])
2.3 编码
2.3.1 编码器
将输入的layer进行重复，然后以前向形式x输入，最后获得的结果进行LayerNorm规范化，并输出
class Encoder(nn.Module):
"Core encoder is a stack of N layers"
def __init__(self, layer, N):
super(Encoder, self).__init__()
self.layers = clones(layer, N)
self.norm = LayerNorm(layer.size)
def forward(self, x, mask):
"Pass the input (and mask) through each layer in turn."
for layer in self.layers:
x = layer(x, mask)
return self.norm(x)
2.3.2 编码层
如图所示，
class EncoderLayer(nn.Module):
"Encoder is made up of self-attn and feed forward (defined below)"
def __init__(self, size, self_attn, feed_forward, dropout):
super(EncoderLayer, self).__init__()
self.self_attn = self_attn
self.feed_forward = feed_forward
self.sublayer = clones(SublayerConnection(size, dropout), 2)
self.size = size
def forward(self, x, mask):
"Follow Figure 1 (left) for connections."
x = self.sublayer[0](x, lambda x: self.self_attn(x, x, x, mask))
return self.sublayer[1](x, self.feed_forward)
2.4 解码
2.4.1 解码器
class Decoder(nn.Module):
"Generic N layer decoder with masking."
def __init__(self, layer, N):
super(Decoder, self).__init__()
self.layers = clones(layer, N)
self.norm = LayerNorm(layer.size)
def forward(self, x, memory, src_mask, tgt_mask):
for layer in self.layers:
x = layer(x, memory, src_mask, tgt_mask)
return self.norm(x)
2.4.2 解码层
class DecoderLayer(nn.Module):
"Decoder is made of self-attn, src-attn, and feed forward (defined below)"
def __init__(self, size, self_attn, src_attn, feed_forward, dropout):
super(DecoderLayer, self).__init__()
self.size = size
self.self_attn = self_attn
# 其实在调用时候，src_attn 就是self_attn的另一个实例，并不是新的attention
self.src_attn = src_attn
self.feed_forward = feed_forward
self.sublayer = clones(SublayerConnection(size, dropout), 3)
def forward(self, x, memory, src_mask, tgt_mask):
"Follow Figure 1 (right) for connections."
m = memory
x = self.sublayer[0](x, lambda x: self.self_attn(x, x, x, tgt_mask))
x = self.sublayer[1](x, lambda x: self.src_attn(x, m, m, src_mask))
return self.sublayer[2](x, self.feed_forward)
2.5 生成器
class Generator(nn.Module):
"Define standard linear + softmax generation step."
def __init__(self, d_model, vocab):
super(Generator, self).__init__()
self.proj = nn.Linear(d_model, vocab)
def forward(self, x):
return F.log_softmax(self.proj(x), dim=-1)
3 MultiHeadedAttention
3.1 前置attention
一个attention函数可以解释成映射一个query和一个kv集合到输出，其中query，keys，values都是向量。输出结果为为values的权重和，其中权重是通过query和对应的key计算得到的，其中queries和keys的维度都为\(d_k\)，values的维度为\(d_v\):
- 1.先计算query和所有keys的点积；
- 2.上述点积结果除以\(\sqrt{d_k}\)
- 3.上述结果输入到softmax获得每个values的权重。
为了一次性同时计算所有的queries，将queries放入矩阵\(Q\),对应的keys和values为\(K\),\(V\):
\[Attention(Q,K,V)=softmax(\frac{QK^T}{\sqrt{d_k}})V
\]
一般而言，用的最多的attention函数无非就是1）加法；2）点积法。其中本文用的就是点积法，只不过其中缩放因子为\(\frac{1}{\sqrt{d_k}}\)。加法attention是使用一个单层隐藏层的feed-forward前向网络计算。这两种其实理论上是相似的，不过点积attention更快，也更省空间，因为它可以调用优化过的矩阵相乘代码。
对于\(d_k\)值较小的情况下，两个机制计算耗时是相似的，加法attention反而由于点积法。我么假设\(d_k\)很大，那么这时候点积值会变得很大,会将softmax函数推到梯度极端小的区域，为了消除这个影响，所以将点积结果乘以\(\frac{1}{\sqrt{d_k}}\)。
def attention(query, key, value, mask=None, dropout=None):
"Compute 'Scaled Dot Product Attention'"
d_k = query.size(-1)
# 1-MatMul & 2-Scale
scores = torch.matmul(query, key.transpose(-2, -1)) \
/ math.sqrt(d_k)
# 3-Mask 可选的
if mask is not None:
scores = scores.masked_fill(mask == 0, -1e9)
# 4-Softmax
p_attn = F.softmax(scores, dim = -1)
if dropout is not None:
p_attn = dropout(p_attn)
# 5-MatMul
return torch.matmul(p_attn, value), p_attn
这里借用《This post is all you need （上卷）——层层剥开Transformer v1.3.1》中的解释，
3.2 MultiHeadedAttention组件
\[MultiHead(Q,K,V) = Concat(head_1,head_2,...,head_n)W_O
\]
其中
\[head_i = Attention(QW_i^Q,KW_i^K,VW_i^V)
\]
这里，
,
其中\(h=8\),并且\(d_k=d_v=\frac{d_{model}}{h}=64\),则
\(W_i^Q\in R^{512*64}\),\(W_i^K\in R^{512*64}\),\(W_i^V\in R^{512*64}\),\(W^O\in R^{512*512}\)
因为减少了每个head的维度，所以总的计算量还是与单个head（全维度）差不多。
class MultiHeadedAttention(nn.Module):
def __init__(self, h, d_model, dropout=0.1):
"Take in model size and number of heads."
super(MultiHeadedAttention, self).__init__()
assert d_model % h == 0
# We assume d_v always equals d_k
# 其中 d_model=64*8, d_k=64, h=8
self.d_k = d_model // h
self.h = h
# 创建4个 64*64的全连接层
self.linears = clones(nn.Linear(d_model, d_model), 4)
self.attn = None
self.dropout = nn.Dropout(p=dropout)
def forward(self, query, key, value, mask=None):
"Implements Figure 2"
if mask is not None:
# Same mask applied to all h heads.
mask = mask.unsqueeze(1)
# 获取当前batch的大小
nbatches = query.size(0)
# 1) Do all the linear projections in batch from d_model => h x d_k
# 通过zip将(query, key, value)与前3个进行一一对应，即[l(query),l(key),l(value)]
query, key, value = [l(x).view(nbatches, -1, self.h, self.d_k ).transpose(1, 2)
for l, x in zip(self.linears, (query, key, value))]
# 2) Apply attention on all the projected vectors in batch.
# 以batch形式计算attention
x, self.attn = attention(query, key, value, mask=mask,
dropout=self.dropout)
# 3) "Concat" using a view and apply a final linear.
# 得到的权重矩阵是 [nbatches,-1,512]
x = x.transpose(1, 2).contiguous() \
.view(nbatches, -1, self.h * self.d_k)
return self.linears[-1](x)
4 PositionwiseFeedForward
逐点相乘
其中输入和输出都是\(d_model=512\),而内层维度是\(d_{ff}=2048\),即\(W_1\in R^{512*2048}\)
class PositionwiseFeedForward(nn.Module):
"Implements FFN equation."
def __init__(self, d_model, d_ff, dropout=0.1):
super(PositionwiseFeedForward, self).__init__()
self.w_1 = nn.Linear(d_model, d_ff)
self.w_2 = nn.Linear(d_ff, d_model)
self.dropout = nn.Dropout(dropout)
def forward(self, x):
return self.w_2(self.dropout(F.relu(self.w_1(x))))
5 PositionalEncoding
5.1 位置编码组件
因为transformer即没rnn，也不是cnn，所以为了让模型获取到序列顺序，就需要将位置的相对信息或者绝对信息注入到 模型中，所以才需要增加"位置编码"到embedding部分，位置编码的维度和embedding一样都是\(d_{model}\)，
本文中采用余弦函数来获取不同位置
；
，其中\(pos\)就是位置，\(i\)是维度，即每个位置的维度信息都对应一个波形，选择这个函数是因为假设它可以让模型很容易学到在基于固定的偏移量\(k\)基础上的相对位置，\(PE_{pos+k}\)可以解释成\(PE_{pos}\)的线性函数。经过实验dropout=0.1时最佳。
class PositionalEncoding(nn.Module):
"Implement the PE function."
def __init__(self, d_model, dropout, max_len=5000):
super(PositionalEncoding, self).__init__()
self.dropout = nn.Dropout(p=dropout)
# Compute the positional encodings once in log space.
pe = torch.zeros(max_len, d_model)
position = torch.arange(0, max_len).unsqueeze(1)
div_term = torch.exp(torch.arange(0, d_model, 2) *
-(math.log(10000.0) / d_model))
pe[:, 0::2] = torch.sin(position * div_term)
pe[:, 1::2] = torch.cos(position * div_term)
pe = pe.unsqueeze(0)
self.register_buffer('pe', pe)
def forward(self, x):
# 输入加上位置信息
# 在实现上，这里x是单词的embedding值，即在embedding上加上位置信息
x = x + Variable(self.pe[:, :x.size(1)],
requires_grad=False)
return self.dropout(x)
5.2 例子
plt.figure(figsize=(15, 5))
pe = PositionalEncoding(20, 0)
y = pe.forward(Variable(torch.zeros(1, 100, 20)))
plt.plot(np.arange(100), y[0, :, 4:8].data.numpy())
plt.legend(["dim %d"%p for p in [4,5,6,7]])
None
6 LabelSmoothing
6.1 组件
标签平滑，虽然会让模型很困惑，因为模型会变得更加不确定，但是能提高准确度和BLEU分数。
主要用于将预测的值进行平滑，并与训练集的真实label求KL散度 loss
class LabelSmoothing(nn.Module):
"Implement label smoothing."
def __init__(self, size, padding_idx, smoothing=0.0):
super(LabelSmoothing, self).__init__()
self.criterion = nn.KLDivLoss(size_average=False)
self.padding_idx = padding_idx
self.confidence = 1.0 - smoothing # 置信度
self.smoothing = smoothing # 平滑系数
self.size = size # 多少列，即一个样本多少类别
self.true_dist = None
def forward(self, x, target):
# x为预测label,target为真实label
# true_dist为平滑后的label，
# 本函数输出的是loss值
assert x.size(1) == self.size
true_dist = x.data.clone() # @1
true_dist.fill_(self.smoothing / (self.size - 2)) # @2
true_dist.scatter_(1, target.data.unsqueeze(1), self.confidence) # @3
true_dist[:, self.padding_idx] = 0 # @4
mask = torch.nonzero(target.data == self.padding_idx) # @5
if mask.dim() > 0:
true_dist.index_fill_(0, mask.squeeze(), 0.0) # @6
self.true_dist = true_dist
return self.criterion(x, Variable(true_dist, requires_grad=False)) # @7
6.2 例子
# Example of label smoothing.
# 5个类别，不padding，平滑系数0.4
crit = LabelSmoothing(5, 0, 0.4)
# 假设模型的预测值为如下：
predict = torch.FloatTensor([[0, 0.2, 0.7, 0.1, 0],
[0, 0.2, 0.7, 0.1, 0],
[0, 0.2, 0.7, 0.1, 0]])
# 获取标签平滑后的loss
v = crit(Variable(predict.log()),
Variable(torch.LongTensor([2, 1, 0])))
# Show the target distributions expected by the system.
plt.imshow(crit.true_dist)
1）@1如下
2）@2如下
3）@3如下
4）@4如下
5）@5如下
6）@6如下
7）@7如下
crit = LabelSmoothing(5, 0, 0.1)
def loss(x):
d = x + 3 * 1
predict = torch.FloatTensor([[0, x / d, 1 / d, 1 / d, 1 / d],
])
#print(predict)
return crit(Variable(predict.log()),
Variable(torch.LongTensor([1]))).data[0]
plt.plot(np.arange(1, 100), [loss(x) for x in range(1, 100)])
7 优化器
7.1 优化器组件
class NoamOpt:
"Optim wrapper that implements rate."
def __init__(self, model_size, factor, warmup, optimizer):
self.optimizer = optimizer
self._step = 0
self.warmup = warmup
self.factor = factor
self.model_size = model_size
self._rate = 0
def step(self):
"Update parameters and rate"
self._step += 1
rate = self.rate()
for p in self.optimizer.param_groups:
p['lr'] = rate
self._rate = rate
self.optimizer.step()
def rate(self, step = None):
"Implement `lrate` above"
if step is None:
step = self._step
return self.factor * \
(self.model_size ** (-0.5) *
min(step ** (-0.5), step * self.warmup ** (-1.5)))
def get_std_opt(model):
return NoamOpt(model.src_embed[0].d_model, 2, 4000,
torch.optim.Adam(model.parameters(), lr=0, betas=(0.9, 0.98), eps=1e-9))
7.2 例子
# Three settings of the lrate hyperparameters.
opts = [NoamOpt(512, 1, 4000, None),
NoamOpt(512, 1, 8000, None),
NoamOpt(256, 1, 4000, None)]
plt.plot(np.arange(1, 20000), [[opt.rate(i) for opt in opts] for i in range(1, 20000)])
plt.legend(["512:4000", "512:8000", "256:4000"])
None
8 训练
8.1 Batch
class Batch:
"Object for holding a batch of data with mask during training."
def __init__(self, src, trg=None, pad=0):
self.src = src
self.src_mask = (src != pad).unsqueeze(-2)
if trg is not None:
self.trg = trg[:, :-1]
self.trg_y = trg[:, 1:]
self.trg_mask = \
self.make_std_mask(self.trg, pad)
self.ntokens = (self.trg_y != pad).data.sum()
@staticmethod
def make_std_mask(tgt, pad):
"Create a mask to hide padding and future words."
tgt_mask = (tgt != pad).unsqueeze(-2)
tgt_mask = tgt_mask & Variable(
subsequent_mask(tgt.size(-1)).type_as(tgt_mask.data))
return tgt_mask
8.2 run_epoch
def run_epoch(data_iter, model, loss_compute):
"Standard Training and Logging Function"
start = time.time()
total_tokens = 0
total_loss = 0
tokens = 0
for i, batch in enumerate(data_iter):
out = model.forward(batch.src, batch.trg,
batch.src_mask, batch.trg_mask)
loss = loss_compute(out, batch.trg_y, batch.ntokens)
total_loss += loss
total_tokens += batch.ntokens
tokens += batch.ntokens
if i % 50 == 1:
elapsed = time.time() - start
print("Epoch Step: %d Loss: %f Tokens per Sec: %f" %
(i, loss / batch.ntokens, tokens / elapsed))
start = time.time()
tokens = 0
return total_loss / total_tokens
8.3 batch_size_fn
global max_src_in_batch, max_tgt_in_batch
def batch_size_fn(new, count, sofar):
"Keep augmenting batch and calculate total number of tokens + padding."
global max_src_in_batch, max_tgt_in_batch
if count == 1:
max_src_in_batch = 0
max_tgt_in_batch = 0
max_src_in_batch = max(max_src_in_batch, len(new.src))
max_tgt_in_batch = max(max_tgt_in_batch, len(new.trg) + 2)
src_elements = count * max_src_in_batch
tgt_elements = count * max_tgt_in_batch
return max(src_elements, tgt_elements)
9 真实世界的例子
9.1 数据装载
# For data loading.
from torchtext import data, datasets
if True:
import spacy
spacy_de = spacy.load('de')
spacy_en = spacy.load('en')
def tokenize_de(text):
return [tok.text for tok in spacy_de.tokenizer(text)]
def tokenize_en(text):
return [tok.text for tok in spacy_en.tokenizer(text)]
BOS_WORD = '<s>'
EOS_WORD = '</s>'
BLANK_WORD = "<blank>"
SRC = data.Field(tokenize=tokenize_de, pad_token=BLANK_WORD)
TGT = data.Field(tokenize=tokenize_en, init_token = BOS_WORD,
eos_token = EOS_WORD, pad_token=BLANK_WORD)
MAX_LEN = 100
train, val, test = datasets.IWSLT.splits(
exts=('.de', '.en'), fields=(SRC, TGT),
filter_pred=lambda x: len(vars(x)['src']) <= MAX_LEN and
len(vars(x)['trg']) <= MAX_LEN)
MIN_FREQ = 2
SRC.build_vocab(train.src, min_freq=MIN_FREQ)
TGT.build_vocab(train.trg, min_freq=MIN_FREQ)
9.2 迭代器
class MyIterator(data.Iterator):
def create_batches(self):
if self.train:
def pool(d, random_shuffler):
for p in data.batch(d, self.batch_size * 100):
p_batch = data.batch(
sorted(p, key=self.sort_key),
self.batch_size, self.batch_size_fn)
for b in random_shuffler(list(p_batch)):
yield b
self.batches = pool(self.data(), self.random_shuffler)
else:
self.batches = []
for b in data.batch(self.data(), self.batch_size,
self.batch_size_fn):
self.batches.append(sorted(b, key=self.sort_key))
def rebatch(pad_idx, batch):
"Fix order in torchtext to match ours"
src, trg = batch.src.transpose(0, 1), batch.trg.transpose(0, 1)
return Batch(src, trg, pad_idx)
9.3 多gpu计算
# Skip if not interested in multigpu.
class MultiGPULossCompute:
"A multi-gpu loss compute and train function."
def __init__(self, generator, criterion, devices, opt=None, chunk_size=5):
# Send out to different gpus.
self.generator = generator
self.criterion = nn.parallel.replicate(criterion,
devices=devices)
self.opt = opt
self.devices = devices
self.chunk_size = chunk_size
def __call__(self, out, targets, normalize):
total = 0.0
generator = nn.parallel.replicate(self.generator,
devices=self.devices)
out_scatter = nn.parallel.scatter(out,
target_gpus=self.devices)
out_grad = [[] for _ in out_scatter]
targets = nn.parallel.scatter(targets,
target_gpus=self.devices)
# Divide generating into chunks.
chunk_size = self.chunk_size
for i in range(0, out_scatter[0].size(1), chunk_size):
# Predict distributions
out_column = [[Variable(o[:, i:i+chunk_size].data,
requires_grad=self.opt is not None)]
for o in out_scatter]
gen = nn.parallel.parallel_apply(generator, out_column)
# Compute loss.
y = [(g.contiguous().view(-1, g.size(-1)),
t[:, i:i+chunk_size].contiguous().view(-1))
for g, t in zip(gen, targets)]
loss = nn.parallel.parallel_apply(self.criterion, y)
# Sum and normalize loss
l = nn.parallel.gather(loss,
target_device=self.devices[0])
l = l.sum()[0] / normalize
total += l.data[0]
# Backprop loss to output of transformer
if self.opt is not None:
l.backward()
for j, l in enumerate(loss):
out_grad[j].append(out_column[j][0].grad.data.clone())
# Backprop all loss through transformer.
if self.opt is not None:
out_grad = [Variable(torch.cat(og, dim=1)) for og in out_grad]
o1 = out
o2 = nn.parallel.gather(out_grad,
target_device=self.devices[0])
o1.backward(gradient=o2)
self.opt.step()
self.opt.optimizer.zero_grad()
return total * normalize
# GPUs to use
devices = [0, 1, 2, 3]
if True:
pad_idx = TGT.vocab.stoi["<blank>"]
model = make_model(len(SRC.vocab), len(TGT.vocab), N=6)
model.cuda()
criterion = LabelSmoothing(size=len(TGT.vocab), padding_idx=pad_idx, smoothing=0.1)
criterion.cuda()
BATCH_SIZE = 12000
train_iter = MyIterator(train, batch_size=BATCH_SIZE, device=0,
repeat=False, sort_key=lambda x: (len(x.src), len(x.trg)),
batch_size_fn=batch_size_fn, train=True)
valid_iter = MyIterator(val, batch_size=BATCH_SIZE, device=0,
repeat=False, sort_key=lambda x: (len(x.src), len(x.trg)),
batch_size_fn=batch_size_fn, train=False)
model_par = nn.DataParallel(model, device_ids=devices)
None
9.4 训练
if False:
model_opt = NoamOpt(model.src_embed[0].d_model, 1, 2000,
torch.optim.Adam(model.parameters(), lr=0, betas=(0.9, 0.98), eps=1e-9))
for epoch in range(10):
model_par.train()
run_epoch((rebatch(pad_idx, b) for b in train_iter),
model_par,
MultiGPULossCompute(model.generator, criterion,
devices=devices, opt=model_opt))
model_par.eval()
loss = run_epoch((rebatch(pad_idx, b) for b in valid_iter),
model_par,
MultiGPULossCompute(model.generator, criterion,
devices=devices, opt=None))
print(loss)
else:
model = torch.load("iwslt.pt")
for i, batch in enumerate(valid_iter):
src = batch.src.transpose(0, 1)[:1]
src_mask = (src != SRC.vocab.stoi["<blank>"]).unsqueeze(-2)
out = greedy_decode(model, src, src_mask,
max_len=60, start_symbol=TGT.vocab.stoi["<s>"])
print("Translation:", end="\t")
for i in range(1, out.size(1)):
sym = TGT.vocab.itos[out[0, i]]
if sym == "</s>": break
print(sym, end =" ")
print()
print("Target:", end="\t")
for i in range(1, batch.trg.size(0)):
sym = TGT.vocab.itos[batch.trg.data[i, 0]]
if sym == "</s>": break
print(sym, end =" ")
print()
break
