This is a tutorial on training a sequence-to-sequence model that uses the
nn.Transformer <https://pytorch.org/docs/stable/generated/torch.nn.Transformer.html>__ module.
The PyTorch 1.2 release includes a standard transformer module based on the
paper Attention is All You Need <https://arxiv.org/pdf/1706.03762.pdf>.
Compared to Recurrent Neural Networks (RNNs), the transformer model has proven
to be superior in quality for many sequence-to-sequence tasks while being more
parallelizable. The nn.Transformer module relies entirely on an attention
mechanism (implemented as
nn.MultiheadAttention <https://pytorch.org/docs/stable/generated/torch.nn.MultiheadAttention.html>)
to draw global dependencies between input and output. The nn.Transformer
module is highly modularized such that a single component (e.g.,
nn.TransformerEncoder <https://pytorch.org/docs/stable/generated/torch.nn.TransformerEncoder.html>__)
can be easily adapted/composed.
Define the model¶
In this tutorial, we train a nn.TransformerEncoder model on a
language modeling task. The language modeling task is to assign a
probability for the likelihood of a given word (or a sequence of words)
to follow a sequence of words. A sequence of tokens are passed to the embedding
layer first, followed by a positional encoding layer to account for the order
of the word (see the next paragraph for more details). The
nn.TransformerEncoder consists of multiple layers of
nn.TransformerEncoderLayer <https://pytorch.org/docs/stable/generated/torch.nn.TransformerEncoderLayer.html>__.
Along with the input sequence, a square attention mask is required because the
self-attention layers in nn.TransformerEncoder are only allowed to attend
the earlier positions in the sequence. For the language modeling task, any
tokens on the future positions should be masked. To produce a probability
distribution over output words, the output of the nn.TransformerEncoder
model is passed through a linear layer followed by a log-softmax function.
import math
from typing import Tuple
import torch
from torch import nn, Tensor
import torch.nn.functional as F
from torch.nn import TransformerEncoder, TransformerEncoderLayer
from torch.utils.data import dataset
class TransformerModel(nn.Module):
def __init__(self, ntoken: int, d_model: int, nhead: int, d_hid: int,
nlayers: int, dropout: float = 0.5):
super().__init__()
self.model_type = 'Transformer'
self.pos_encoder = PositionalEncoding(d_model, dropout)
encoder_layers = TransformerEncoderLayer(d_model, nhead, d_hid, dropout)
self.transformer_encoder = TransformerEncoder(encoder_layers, nlayers)
self.encoder = nn.Embedding(ntoken, d_model)
self.d_model = d_model
self.decoder = nn.Linear(d_model, ntoken)
self.init_weights()
def init_weights(self) -> None:
initrange = 0.1
self.encoder.weight.data.uniform_(-initrange, initrange)
self.decoder.bias.data.zero_()
self.decoder.weight.data.uniform_(-initrange, initrange)
def forward(self, src: Tensor, src_mask: Tensor) -> Tensor:
"""
Args:
src: Tensor, shape [seq_len, batch_size]
src_mask: Tensor, shape [seq_len, seq_len]
Returns:
output Tensor of shape [seq_len, batch_size, ntoken]
"""
src = self.encoder(src) * math.sqrt(self.d_model)
src = self.pos_encoder(src)
output = self.transformer_encoder(src, src_mask)
output = self.decoder(output)
return output
def generate_square_subsequent_mask(sz: int) -> Tensor:
"""Generates an upper-triangular matrix of -inf, with zeros on diag."""
return torch.triu(torch.ones(sz, sz) * float('-inf'), diagonal=1)
PositionalEncoding module injects some information about the
relative or absolute position of the tokens in the sequence. The
positional encodings have the same dimension as the embeddings so that
the two can be summed. Here, we use sine and cosine functions of
different frequencies.
class PositionalEncoding(nn.Module):
def __init__(self, d_model: int, dropout: float = 0.1, max_len: int = 5000):
super().__init__()
self.dropout = nn.Dropout(p=dropout)
position = torch.arange(max_len).unsqueeze(1)
div_term = torch.exp(torch.arange(0, d_model, 2) * (-math.log(10000.0) / d_model))
pe = torch.zeros(max_len, 1, d_model)
pe[:, 0, 0::2] = torch.sin(position * div_term)
pe[:, 0, 1::2] = torch.cos(position * div_term)
self.register_buffer('pe', pe)
def forward(self, x: Tensor) -> Tensor:
"""
Args:
x: Tensor, shape [seq_len, batch_size, embedding_dim]
"""
x = x + self.pe[:x.size(0)]
return self.dropout(x)
Load and batch data¶
This tutorial uses torchtext to generate Wikitext-2 dataset.
To access torchtext datasets, please install torchdata following instructions at https://github.com/pytorch/data.
The vocab object is built based on the train dataset and is used to numericalize
tokens into tensors. Wikitext-2 represents rare tokens as <unk>.
Given a 1-D vector of sequential data, batchify() arranges the data
into batch_size columns. If the data does not divide evenly into
batch_size columns, then the data is trimmed to fit. For instance, with
the alphabet as the data (total length of 26) and batch_size=4, we would
divide the alphabet into 4 sequences of length 6:
Batching enables more parallelizable processing. However, batching means that
the model treats each column independently; for example, the dependence of
G and F can not be learned in the example above.
from torchtext.datasets import WikiText2
from torchtext.data.utils import get_tokenizer
from torchtext.vocab import build_vocab_from_iterator
train_iter = WikiText2(split='train')
tokenizer = get_tokenizer('basic_english')
vocab = build_vocab_from_iterator(map(tokenizer, train_iter), specials=['<unk>'])
vocab.set_default_index(vocab['<unk>'])
def data_process(raw_text_iter: dataset.IterableDataset) -> Tensor:
"""Converts raw text into a flat Tensor."""
data = [torch.tensor(vocab(tokenizer(item)), dtype=torch.long) for item in raw_text_iter]
return torch.cat(tuple(filter(lambda t: t.numel() > 0, data)))
# train_iter was "consumed" by the process of building the vocab,
# so we have to create it again
train_iter, val_iter, test_iter = WikiText2()
train_data = data_process(train_iter)
val_data = data_process(val_iter)
test_data = data_process(test_iter)
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
def batchify(data: Tensor, bsz: int) -> Tensor:
"""Divides the data into bsz separate sequences, removing extra elements
that wouldn't cleanly fit.
Args:
data: Tensor, shape [N]
bsz: int, batch size
Returns:
Tensor of shape [N // bsz, bsz]
"""
seq_len = data.size(0) // bsz
data = data[:seq_len * bsz]
data = data.view(bsz, seq_len).t().contiguous()
return data.to(device)
batch_size = 20
eval_batch_size = 10
train_data = batchify(train_data, batch_size) # shape [seq_len, batch_size]
val_data = batchify(val_data, eval_batch_size)
test_data = batchify(test_data, eval_batch_size)
Functions to generate input and target sequence
~~~~~~~~~~~~~
get_batch() generates a pair of input-target sequences for
the transformer model. It subdivides the source data into chunks of
length bptt. For the language modeling task, the model needs the
following words as Target. For example, with a bptt value of 2,
we’d get the following two Variables for i = 0:
It should be noted that the chunks are along dimension 0, consistent
with the S dimension in the Transformer model. The batch dimension
N is along dimension 1.
bptt = 35
def get_batch(source: Tensor, i: int) -> Tuple[Tensor, Tensor]:
"""
Args:
source: Tensor, shape [full_seq_len, batch_size]
i: int
Returns:
tuple (data, target), where data has shape [seq_len, batch_size] and
target has shape [seq_len * batch_size]
"""
seq_len = min(bptt, len(source) - 1 - i)
data = source[i:i+seq_len]
target = source[i+1:i+1+seq_len].reshape(-1)
return data, target
Initiate an instance¶
The model hyperparameters are defined below. The vocab size is equal to the length of the vocab object.
ntokens = len(vocab) # size of vocabulary
emsize = 200 # embedding dimension
d_hid = 200 # dimension of the feedforward network model in nn.TransformerEncoder
nlayers = 2 # number of nn.TransformerEncoderLayer in nn.TransformerEncoder
nhead = 2 # number of heads in nn.MultiheadAttention
dropout = 0.2 # dropout probability
model = TransformerModel(ntokens, emsize, nhead, d_hid, nlayers, dropout).to(device)
Run the model¶
We use CrossEntropyLoss <https://pytorch.org/docs/stable/generated/torch.nn.CrossEntropyLoss.html>
with the SGD <https://pytorch.org/docs/stable/generated/torch.optim.SGD.html>
(stochastic gradient descent) optimizer. The learning rate is initially set to
5.0 and follows a StepLR <https://pytorch.org/docs/stable/generated/torch.optim.lr_scheduler.StepLR.html>
schedule. During training, we use nn.utils.clip_grad_norm\_ <https://pytorch.org/docs/stable/generated/torch.nn.utils.clip_grad_norm_.html>
to prevent gradients from exploding.
import copy
import time
criterion = nn.CrossEntropyLoss()
lr = 5.0 # learning rate
optimizer = torch.optim.SGD(model.parameters(), lr=lr)
scheduler = torch.optim.lr_scheduler.StepLR(optimizer, 1.0, gamma=0.95)
def train(model: nn.Module) -> None:
model.train() # turn on train mode
total_loss = 0.
log_interval = 200
start_time = time.time()
src_mask = generate_square_subsequent_mask(bptt).to(device)
num_batches = len(train_data) // bptt
for batch, i in enumerate(range(0, train_data.size(0) - 1, bptt)):
data, targets = get_batch(train_data, i)
batch_size = data.size(0)
if batch_size != bptt: # only on last batch
src_mask = src_mask[:batch_size, :batch_size]
output = model(data, src_mask)
loss = criterion(output.view(-1, ntokens), targets)
optimizer.zero_grad()
loss.backward()
torch.nn.utils.clip_grad_norm_(model.parameters(), 0.5)
optimizer.step()
total_loss += loss.item()
if batch % log_interval == 0 and batch > 0:
lr = scheduler.get_last_lr()[0]
ms_per_batch = (time.time() - start_time) * 1000 / log_interval
cur_loss = total_loss / log_interval
ppl = math.exp(cur_loss)
print(f'| epoch {epoch:3d} | {batch:5d}/{num_batches:5d} batches | '
f'lr {lr:02.2f} | ms/batch {ms_per_batch:5.2f} | '
f'loss {cur_loss:5.2f} | ppl {ppl:8.2f}')
total_loss = 0
start_time = time.time()
def evaluate(model: nn.Module, eval_data: Tensor) -> float:
model.eval() # turn on evaluation mode
total_loss = 0.
src_mask = generate_square_subsequent_mask(bptt).to(device)
with torch.no_grad():
for i in range(0, eval_data.size(0) - 1, bptt):
data, targets = get_batch(eval_data, i)
batch_size = data.size(0)
if batch_size != bptt:
src_mask = src_mask[:batch_size, :batch_size]
output = model(data, src_mask)
output_flat = output.view(-1, ntokens)
total_loss += batch_size * criterion(output_flat, targets).item()
return total_loss / (len(eval_data) - 1)
Loop over epochs. Save the model if the validation loss is the best we've seen so far. Adjust the learning rate after each epoch.
best_val_loss = float('inf')
epochs = 3
best_model = None
for epoch in range(1, epochs + 1):
epoch_start_time = time.time()
train(model)
val_loss = evaluate(model, val_data)
val_ppl = math.exp(val_loss)
elapsed = time.time() - epoch_start_time
print('-' * 89)
print(f'| end of epoch {epoch:3d} | time: {elapsed:5.2f}s | '
f'valid loss {val_loss:5.2f} | valid ppl {val_ppl:8.2f}')
print('-' * 89)
if val_loss < best_val_loss:
best_val_loss = val_loss
best_model = copy.deepcopy(model)
scheduler.step()
Evaluate the best model on the test dataset¶
test_loss = evaluate(best_model, test_data)
test_ppl = math.exp(test_loss)
print('=' * 89)
print(f'| End of training | test loss {test_loss:5.2f} | '
f'test ppl {test_ppl:8.2f}')
print('=' * 89)