” bitiştirilmedir. "BERT girdi
dizisi" terminolojisini diğer "diziler" türlerinden sürekli olarak ayırt
edeceğiz. Örneğin, bir *BERT girdi dizisi*, bir *metin dizisi* veya iki
*metin dizisi* içerebilir.
Metin çiftlerini ayırt etmek için, öğrenilmiş bölüm gömmeleri
:math:`\mathbf{e}_A` ve :math:`\mathbf{e}_B` sırasıyla birinci dizinin
ve ikinci dizinin belirteç gömmelerine eklenir. Tek metin girdileri için
sadece :math:`\mathbf{e}_A` kullanılır.
Aşağıdaki ``get_tokens_and_segments`` girdi olarak bir veya iki cümle
alır, daha sonra BERT girdi dizisinin belirteçlerini ve bunlara karşılık
gelen bölüm kimliklerini döndürür.
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
#@save
def get_tokens_and_segments(tokens_a, tokens_b=None):
"""BERT girdi dizisinin belirteçlerini ve bunların bölüm kimliklerini alın."""
tokens = [''] + tokens_a + ['']
# 0 and 1 are marking segment A and B, respectively
segments = [0] * (len(tokens_a) + 2)
if tokens_b is not None:
tokens += tokens_b + ['']
segments += [1] * (len(tokens_b) + 1)
return tokens, segments
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
#@save
def get_tokens_and_segments(tokens_a, tokens_b=None):
"""BERT girdi dizisinin belirteçlerini ve bunların bölüm kimliklerini alın."""
tokens = [''] + tokens_a + ['']
# 0 and 1 are marking segment A and B, respectively
segments = [0] * (len(tokens_a) + 2)
if tokens_b is not None:
tokens += tokens_b + ['']
segments += [1] * (len(tokens_b) + 1)
return tokens, segments
.. raw:: html
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
#@save
class BERTEncoder(nn.Block):
"""BERT kodlayıcı."""
def __init__(self, vocab_size, num_hiddens, ffn_num_hiddens, num_heads,
num_layers, dropout, max_len=1000, **kwargs):
super(BERTEncoder, self).__init__(**kwargs)
self.token_embedding = nn.Embedding(vocab_size, num_hiddens)
self.segment_embedding = nn.Embedding(2, num_hiddens)
self.blks = nn.Sequential()
for _ in range(num_layers):
self.blks.add(d2l.EncoderBlock(
num_hiddens, ffn_num_hiddens, num_heads, dropout, True))
# BERT'te konumsal gömmeler öğrenilebilir, bu nedenle yeterince
# uzun bir konumsal gömme parametresi oluşturuyoruz
self.pos_embedding = self.params.get('pos_embedding',
shape=(1, max_len, num_hiddens))
def forward(self, tokens, segments, valid_lens):
# Aşağıdaki kod parçacığında `X`'in şekli değişmeden kalır:
# (parti boyutu, maksimum dizi uzunluğu, `num_hiddens`)
X = self.token_embedding(tokens) + self.segment_embedding(segments)
X = X + self.pos_embedding.data(ctx=X.ctx)[:, :X.shape[1], :]
for blk in self.blks:
X = blk(X, valid_lens)
return X
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
#@save
class BERTEncoder(nn.Module):
"""BERT kodlayıcı."""
def __init__(self, vocab_size, num_hiddens, norm_shape, ffn_num_input,
ffn_num_hiddens, num_heads, num_layers, dropout,
max_len=1000, key_size=768, query_size=768, value_size=768,
**kwargs):
super(BERTEncoder, self).__init__(**kwargs)
self.token_embedding = nn.Embedding(vocab_size, num_hiddens)
self.segment_embedding = nn.Embedding(2, num_hiddens)
self.blks = nn.Sequential()
for i in range(num_layers):
self.blks.add_module(f"{i}", d2l.EncoderBlock(
key_size, query_size, value_size, num_hiddens, norm_shape,
ffn_num_input, ffn_num_hiddens, num_heads, dropout, True))
# BERT'te konumsal gömmeler öğrenilebilir, bu nedenle yeterince
# uzun bir konumsal gömme parametresi oluşturuyoruz
self.pos_embedding = nn.Parameter(torch.randn(1, max_len,
num_hiddens))
def forward(self, tokens, segments, valid_lens):
# Aşağıdaki kod parçacığında `X`'in şekli değişmeden kalır:
# (parti boyutu, maksimum dizi uzunluğu, `num_hiddens`)
X = self.token_embedding(tokens) + self.segment_embedding(segments)
X = X + self.pos_embedding.data[:, :X.shape[1], :]
for blk in self.blks:
X = blk(X, valid_lens)
return X
.. raw:: html
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
vocab_size, num_hiddens, ffn_num_hiddens, num_heads = 10000, 768, 1024, 4
num_layers, dropout = 2, 0.2
encoder = BERTEncoder(vocab_size, num_hiddens, ffn_num_hiddens, num_heads,
num_layers, dropout)
encoder.initialize()
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
vocab_size, num_hiddens, ffn_num_hiddens, num_heads = 10000, 768, 1024, 4
norm_shape, ffn_num_input, num_layers, dropout = [768], 768, 2, 0.2
encoder = BERTEncoder(vocab_size, num_hiddens, norm_shape, ffn_num_input,
ffn_num_hiddens, num_heads, num_layers, dropout)
.. raw:: html
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
tokens = np.random.randint(0, vocab_size, (2, 8))
segments = np.array([[0, 0, 0, 0, 1, 1, 1, 1], [0, 0, 0, 1, 1, 1, 1, 1]])
encoded_X = encoder(tokens, segments, None)
encoded_X.shape
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
(2, 8, 768)
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
tokens = torch.randint(0, vocab_size, (2, 8))
segments = torch.tensor([[0, 0, 0, 0, 1, 1, 1, 1], [0, 0, 0, 1, 1, 1, 1, 1]])
encoded_X = encoder(tokens, segments, None)
encoded_X.shape
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
torch.Size([2, 8, 768])
.. raw:: html
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
#@save
class MaskLM(nn.Block):
"""BERT'in maskeli dil modeli görevi."""
def __init__(self, vocab_size, num_hiddens, **kwargs):
super(MaskLM, self).__init__(**kwargs)
self.mlp = nn.Sequential()
self.mlp.add(
nn.Dense(num_hiddens, flatten=False, activation='relu'))
self.mlp.add(nn.LayerNorm())
self.mlp.add(nn.Dense(vocab_size, flatten=False))
def forward(self, X, pred_positions):
num_pred_positions = pred_positions.shape[1]
pred_positions = pred_positions.reshape(-1)
batch_size = X.shape[0]
batch_idx = np.arange(0, batch_size)
# Varsayalım ki `batch_size` = 2, `num_pred_positions` = 3 olsun,
# o halde `batch_idx` `np.array([0, 0, 0, 1, 1, 1])` olur
batch_idx = np.repeat(batch_idx, num_pred_positions)
masked_X = X[batch_idx, pred_positions]
masked_X = masked_X.reshape((batch_size, num_pred_positions, -1))
mlm_Y_hat = self.mlp(masked_X)
return mlm_Y_hat
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
#@save
class MaskLM(nn.Module):
"""BERT'in maskeli dil modeli görevi."""
def __init__(self, vocab_size, num_hiddens, num_inputs=768, **kwargs):
super(MaskLM, self).__init__(**kwargs)
self.mlp = nn.Sequential(nn.Linear(num_inputs, num_hiddens),
nn.ReLU(),
nn.LayerNorm(num_hiddens),
nn.Linear(num_hiddens, vocab_size))
def forward(self, X, pred_positions):
num_pred_positions = pred_positions.shape[1]
pred_positions = pred_positions.reshape(-1)
batch_size = X.shape[0]
batch_idx = torch.arange(0, batch_size)
# Varsayalım ki `batch_size` = 2, `num_pred_positions` = 3 olsun,
# o halde `batch_idx` `np.array([0, 0, 0, 1, 1, 1])` olur
batch_idx = torch.repeat_interleave(batch_idx, num_pred_positions)
masked_X = X[batch_idx, pred_positions]
masked_X = masked_X.reshape((batch_size, num_pred_positions, -1))
mlm_Y_hat = self.mlp(masked_X)
return mlm_Y_hat
.. raw:: html
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
mlm = MaskLM(vocab_size, num_hiddens)
mlm.initialize()
mlm_positions = np.array([[1, 5, 2], [6, 1, 5]])
mlm_Y_hat = mlm(encoded_X, mlm_positions)
mlm_Y_hat.shape
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
(2, 3, 10000)
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
mlm = MaskLM(vocab_size, num_hiddens)
mlm_positions = torch.tensor([[1, 5, 2], [6, 1, 5]])
mlm_Y_hat = mlm(encoded_X, mlm_positions)
mlm_Y_hat.shape
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
torch.Size([2, 3, 10000])
.. raw:: html
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
mlm_Y = np.array([[7, 8, 9], [10, 20, 30]])
loss = gluon.loss.SoftmaxCrossEntropyLoss()
mlm_l = loss(mlm_Y_hat.reshape((-1, vocab_size)), mlm_Y.reshape(-1))
mlm_l.shape
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
(6,)
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
mlm_Y = torch.tensor([[7, 8, 9], [10, 20, 30]])
loss = nn.CrossEntropyLoss(reduction='none')
mlm_l = loss(mlm_Y_hat.reshape((-1, vocab_size)), mlm_Y.reshape(-1))
mlm_l.shape
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
torch.Size([6])
.. raw:: html
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
#@save
class NextSentencePred(nn.Block):
"""BERT'in sonraki cümle tahmini görevi."""
def __init__(self, **kwargs):
super(NextSentencePred, self).__init__(**kwargs)
self.output = nn.Dense(2)
def forward(self, X):
# `X` shape: (batch size, `num_hiddens`)
return self.output(X)
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
#@save
class NextSentencePred(nn.Module):
"""BERT'in sonraki cümle tahmini görevi."""
def __init__(self, num_inputs, **kwargs):
super(NextSentencePred, self).__init__(**kwargs)
self.output = nn.Linear(num_inputs, 2)
def forward(self, X):
# `X` shape: (batch size, `num_hiddens`)
return self.output(X)
.. raw:: html
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
nsp = NextSentencePred()
nsp.initialize()
nsp_Y_hat = nsp(encoded_X)
nsp_Y_hat.shape
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
(2, 2)
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
# PyTorch by default won't flatten the tensor as seen in mxnet where, if
# flatten=True, all but the first axis of input data are collapsed together
encoded_X = torch.flatten(encoded_X, start_dim=1)
# input_shape for NSP: (batch size, `num_hiddens`)
nsp = NextSentencePred(encoded_X.shape[-1])
nsp_Y_hat = nsp(encoded_X)
nsp_Y_hat.shape
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
torch.Size([2, 2])
.. raw:: html
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
nsp_y = np.array([0, 1])
nsp_l = loss(nsp_Y_hat, nsp_y)
nsp_l.shape
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
(2,)
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
nsp_y = torch.tensor([0, 1])
nsp_l = loss(nsp_Y_hat, nsp_y)
nsp_l.shape
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
torch.Size([2])
.. raw:: html
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
#@save
class BERTModel(nn.Block):
"""BERT modeli."""
def __init__(self, vocab_size, num_hiddens, ffn_num_hiddens, num_heads,
num_layers, dropout, max_len=1000):
super(BERTModel, self).__init__()
self.encoder = BERTEncoder(vocab_size, num_hiddens, ffn_num_hiddens,
num_heads, num_layers, dropout, max_len)
self.hidden = nn.Dense(num_hiddens, activation='tanh')
self.mlm = MaskLM(vocab_size, num_hiddens)
self.nsp = NextSentencePred()
def forward(self, tokens, segments, valid_lens=None, pred_positions=None):
encoded_X = self.encoder(tokens, segments, valid_lens)
if pred_positions is not None:
mlm_Y_hat = self.mlm(encoded_X, pred_positions)
else:
mlm_Y_hat = None
# Bir sonraki cümle tahmini için MLP sınıflandırıcısının gizli katmanı.
# 0, '' belirtecinin dizinidir
nsp_Y_hat = self.nsp(self.hidden(encoded_X[:, 0, :]))
return encoded_X, mlm_Y_hat, nsp_Y_hat
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
#@save
class BERTModel(nn.Module):
"""BERT modeli."""
def __init__(self, vocab_size, num_hiddens, norm_shape, ffn_num_input,
ffn_num_hiddens, num_heads, num_layers, dropout,
max_len=1000, key_size=768, query_size=768, value_size=768,
hid_in_features=768, mlm_in_features=768,
nsp_in_features=768):
super(BERTModel, self).__init__()
self.encoder = BERTEncoder(vocab_size, num_hiddens, norm_shape,
ffn_num_input, ffn_num_hiddens, num_heads, num_layers,
dropout, max_len=max_len, key_size=key_size,
query_size=query_size, value_size=value_size)
self.hidden = nn.Sequential(nn.Linear(hid_in_features, num_hiddens),
nn.Tanh())
self.mlm = MaskLM(vocab_size, num_hiddens, mlm_in_features)
self.nsp = NextSentencePred(nsp_in_features)
def forward(self, tokens, segments, valid_lens=None, pred_positions=None):
encoded_X = self.encoder(tokens, segments, valid_lens)
if pred_positions is not None:
mlm_Y_hat = self.mlm(encoded_X, pred_positions)
else:
mlm_Y_hat = None
# Bir sonraki cümle tahmini için MLP sınıflandırıcısının gizli katmanı.
# 0, '' belirtecinin dizinidir
nsp_Y_hat = self.nsp(self.hidden(encoded_X[:, 0, :]))
return encoded_X, mlm_Y_hat, nsp_Y_hat
.. raw:: html
.. raw:: html
.. raw:: html
`Tartışmalar `__
.. raw:: html
.. raw:: html
`Tartışmalar `__
.. raw:: html
.. raw:: html