在15.4 节中,我们在一个小数据集上训练了一个 word2vec 模型,并将其应用于为输入词寻找语义相似的词。在实践中,在大型语料库上预训练的词向量可以应用于下游的自然语言处理任务,这将在第 16 节后面介绍。为了以直接的方式展示来自大型语料库的预训练词向量的语义,让我们将它们应用到词相似度和类比任务中。
15.7.1。加载预训练词向量
下面列出了维度为 50、100 和 300 的预训练 GloVe 嵌入,可以从GloVe 网站下载。预训练的 fastText 嵌入有多种语言版本。这里我们考虑一个可以从fastText 网站下载的英文版本(300 维“wiki.en”) 。
#@save
d2l.DATA_HUB['glove.6b.50d'] = (d2l.DATA_URL + 'glove.6B.50d.zip',
'0b8703943ccdb6eb788e6f091b8946e82231bc4d')
#@save
d2l.DATA_HUB['glove.6b.100d'] = (d2l.DATA_URL + 'glove.6B.100d.zip',
'cd43bfb07e44e6f27cbcc7bc9ae3d80284fdaf5a')
#@save
d2l.DATA_HUB['glove.42b.300d'] = (d2l.DATA_URL + 'glove.42B.300d.zip',
'b5116e234e9eb9076672cfeabf5469f3eec904fa')
#@save
d2l.DATA_HUB['wiki.en'] = (d2l.DATA_URL + 'wiki.en.zip',
'c1816da3821ae9f43899be655002f6c723e91b88')
#@save
d2l.DATA_HUB['glove.6b.50d'] = (d2l.DATA_URL + 'glove.6B.50d.zip',
'0b8703943ccdb6eb788e6f091b8946e82231bc4d')
#@save
d2l.DATA_HUB['glove.6b.100d'] = (d2l.DATA_URL + 'glove.6B.100d.zip',
'cd43bfb07e44e6f27cbcc7bc9ae3d80284fdaf5a')
#@save
d2l.DATA_HUB['glove.42b.300d'] = (d2l.DATA_URL + 'glove.42B.300d.zip',
'b5116e234e9eb9076672cfeabf5469f3eec904fa')
#@save
d2l.DATA_HUB['wiki.en'] = (d2l.DATA_URL + 'wiki.en.zip',
'c1816da3821ae9f43899be655002f6c723e91b88')
为了加载这些预训练的 GloVe 和 fastText 嵌入,我们定义了以下TokenEmbedding
类。
#@save
class TokenEmbedding:
"""Token Embedding."""
def __init__(self, embedding_name):
self.idx_to_token, self.idx_to_vec = self._load_embedding(
embedding_name)
self.unknown_idx = 0
self.token_to_idx = {token: idx for idx, token in
enumerate(self.idx_to_token)}
def _load_embedding(self, embedding_name):
idx_to_token, idx_to_vec = [''], []
data_dir = d2l.download_extract(embedding_name)
# GloVe website: https://nlp.stanford.edu/projects/glove/
# fastText website: https://fasttext.cc/
with open(os.path.join(data_dir, 'vec.txt'), 'r') as f:
for line in f:
elems = line.rstrip().split(' ')
token, elems = elems[0], [float(elem) for elem in elems[1:]]
# Skip header information, such as the top row in fastText
if len(elems) > 1:
idx_to_token.append(token)
idx_to_vec.append(elems)
idx_to_vec = [[0] * len(idx_to_vec[0])] + idx_to_vec
return idx_to_token, torch.tensor(idx_to_vec)
def __getitem__(self, tokens):
indices = [self.token_to_idx.get(token, self.unknown_idx)
for token in tokens]
vecs = self.idx_to_vec[torch.tensor(indices)]
return vecs
def __len__(self):
return len(self.idx_to_token)
#@save
class TokenEmbedding:
"""Token Embedding."""
def __init__(self, embedding_name):
self.idx_to_token, self.idx_to_vec = self._load_embedding(
embedding_name)
self.unknown_idx = 0
self.token_to_idx = {token: idx for idx, token in
enumerate(self.idx_to_token)}
def _load_embedding(self, embedding_name):
idx_to_token, idx_to_vec = [''], []
data_dir = d2l.download_extract(embedding_name)
# GloVe website: https://nlp.stanford.edu/projects/glove/
# fastText website: https://fasttext.cc/
with open(os.path.join(data_dir, 'vec.txt'), 'r') as f:
for line in f:
elems = line.rstrip().split(' ')
token, elems = elems[0], [float(elem) for elem in elems[1:]]
# Skip header information, such as the top row in fastText
if len(elems) > 1:
idx_to_token.append(token)
idx_to_vec.append(elems)
idx_to_vec = [[0] * len(idx_to_vec[0])] + idx_to_vec
return idx_to_token, np.array(idx_to_vec)
def __getitem__(self, tokens):
indices = [self.token_to_idx.get(token, self.unknown_idx)
for token in tokens]
vecs = self.idx_to_vec[np.array(indices)]
return vecs
def __len__(self):
return len(self.idx_to_token)
下面我们加载 50 维 GloVe 嵌入(在维基百科子集上预训练)。创建TokenEmbedding
实例时,如果尚未下载指定的嵌入文件,则必须下载。
Downloading ../data/glove.6B.50d.zip from http://d2l-data.s3-accelerate.amazonaws.com/glove.6B.50d.zip...
输出词汇量。词汇表包含 400000 个单词(标记)和一个特殊的未知标记。
我们可以获得一个词在词汇表中的索引,反之亦然。
15.7.2。应用预训练词向量
使用加载的 GloVe 向量,我们将通过将它们应用于以下单词相似性和类比任务来演示它们的语义。
15.7.2.1。词相似度
与第 15.4.3 节类似,为了根据词向量之间的余弦相似度为输入词找到语义相似的词,我们实现以下knn
(k-最近的邻居)功能。
def knn(W, x, k):
# Add 1e-9 for numerical stability
cos = np.dot(W, x.reshape(-1,)) / (
np.sqrt(np.sum(W * W, axis=1) + 1e-9) * np