explosion/spaCy
1
1
Fork 0
mirror of https://github.com/explosion/spaCy.git synced 2025-01-13 18:56:36 +03:00
Code Issues Packages Projects Releases Wiki Activity

Fix embedding in chainer sentiment example

Browse Source
This commit is contained in:
Matthew Honnibal 2016-11-19 19:05:37 +01:00
parent 8a2de46fcb
commit b701a08249
1 changed files with 10 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
examples/chainer_sentiment.py
View File

@ -64,6 +64,7 @@ class SentimentAnalyser(object):
# For arbitrary data storage, there's: # For arbitrary data storage, there's:
# doc.user_data['my_data'] = y # doc.user_data['my_data'] = y
class Classifier(Chain): class Classifier(Chain):
def __init__(self, predictor): def __init__(self, predictor):
super(Classifier, self).__init__(predictor=predictor) super(Classifier, self).__init__(predictor=predictor)
@ -77,9 +78,10 @@ class Classifier(Chain):
class SentimentModel(Chain): class SentimentModel(Chain):
def __init__(self, shape, **settings): def __init__(self, nlp, shape, **settings):
Chain.__init__(self, Chain.__init__(self,
embed=_Embed(shape['nr_vector'], shape['nr_dim'], shape['nr_hidden']), embed=_Embed(shape['nr_vector'], shape['nr_dim'], shape['nr_hidden'],
initialW=lambda arr: set_vectors(arr, nlp.vocab)),
encode=_Encode(shape['nr_hidden'], shape['nr_hidden']), encode=_Encode(shape['nr_hidden'], shape['nr_hidden']),
attend=_Attend(shape['nr_hidden'], shape['nr_hidden']), attend=_Attend(shape['nr_hidden'], shape['nr_hidden']),
predict=_Predict(shape['nr_hidden'], shape['nr_class'])) predict=_Predict(shape['nr_hidden'], shape['nr_class']))
@ -205,16 +207,14 @@ def get_features(docs, max_length):
return Xs return Xs
def get_embeddings(vocab, max_rank=1000): def set_vectors(vectors, vocab):
if max_rank is None:
max_rank = max(lex.rank+1 for lex in vocab if lex.has_vector)
vectors = xp.ndarray((max_rank+1, vocab.vectors_length), dtype='f')
for lex in vocab: for lex in vocab:
if lex.has_vector and lex.rank < max_rank: if lex.has_vector and (lex.rank+1) < vectors.shape[0]:
lex.norm = lex.rank+1 lex.norm = lex.rank+1
vectors[lex.rank + 1] = lex.vector vectors[lex.rank + 1] = lex.vector
else: else:
lex.norm = 0 lex.norm = 0
vectors.unchain_backwards()
return vectors return vectors
@ -222,13 +222,10 @@ def train(train_texts, train_labels, dev_texts, dev_labels,
lstm_shape, lstm_settings, lstm_optimizer, batch_size=100, nb_epoch=5, lstm_shape, lstm_settings, lstm_optimizer, batch_size=100, nb_epoch=5,
by_sentence=True): by_sentence=True):
nlp = spacy.load('en', entity=False) nlp = spacy.load('en', entity=False)
for lex in nlp.vocab: if 'nr_vector' not in lstm_shape:
if lex.rank >= (lstm_shape['nr_vector'] - 1): lstm_shape['nr_vector'] = max(lex.rank+1 for lex in vocab if lex.has_vector)
lex.norm = 0
else:
lex.norm = lex.rank+1
print("Make model") print("Make model")
model = Classifier(SentimentModel(lstm_shape, **lstm_settings)) model = Classifier(SentimentModel(nlp, lstm_shape, **lstm_settings))
print("Parsing texts...") print("Parsing texts...")
if by_sentence: if by_sentence:
train_data = SentenceDataset(nlp, train_texts, train_labels, lstm_shape['max_length']) train_data = SentenceDataset(nlp, train_texts, train_labels, lstm_shape['max_length'])