def _create_char_features(self, sentences, sentence_length, word_length):
char_dict = {}
char_id = 3
new_sentences = []
for s in sentences:
char_sents = []
for w in s:
char_vector = []
for c in w:
char_int = char_dict.get(c, None)
if char_int is None:
char_dict[c] = char_id
char_int = char_id
char_id += 1
char_vector.append(char_int)
char_vector = [1] + char_vector + [2]
char_sents.append(char_vector)
char_sents = pad_sentences(char_sents, sentence_length=word_length)
if sentence_length - char_sents.shape[0] < 0:
char_sents = char_sents[:sentence_length]
else:
padding = np.zeros(
(sentence_length - char_sents.shape[0], word_length))
char_sents = np.vstack((padding, char_sents))
new_sentences.append(char_sents)
char_sentences = np.asarray(new_sentences)
self.vocabs.update({'char_rnn': char_dict})
return char_sentences
def _create_char_features(self, sentences, sentence_length, word_length):
char_dict = {}
char_id = 3
new_sentences = []
for s in sentences:
char_sents = []
for w in s:
char_vector = []
for c in w:
char_int = char_dict.get(c, None)
if char_int is None:
char_dict[c] = char_id
char_int = char_id
char_id += 1
char_vector.append(char_int)
char_vector = [1] + char_vector + [2]
char_sents.append(char_vector)
char_sents = pad_sentences(char_sents, sentence_length=word_length)
if sentence_length - char_sents.shape[0] < 0:
char_sents = char_sents[:sentence_length]
else:
padding = np.zeros(
(sentence_length - char_sents.shape[0], word_length))
char_sents = np.vstack((padding, char_sents))
new_sentences.append(char_sents)
char_sentences = np.asarray(new_sentences)
self.vocabs.update({'char_rnn': char_dict})
return char_sentences
def pad_sentences(sentences, sentence_length=None, dtype=np.int32, pad_val=0.):
logger.error('pad_sentances in the Text class is deprecated. This function'
'is now in neon.data.text_preprocessing.')
return pad_sentences(sentences,
sentence_length=sentence_length,
dtype=dtype,
pad_val=pad_val)
def vectorize_stories(self, data):
"""
Convert (story, query, answer) word data into vectors.
Args:
data (tuple) : Tuple of story, query, answer word data.
Returns:
tuple : Tuple of story, query, answer vectors.
"""
s, q, a = [], [], []
for story, query, answer in data:
s.append(self.words_to_vector(story))
q.append(self.words_to_vector(query))
a.append(self.one_hot_vector(answer))
s = pad_sentences(s, self.story_maxlen)
q = pad_sentences(q, self.query_maxlen)
a = np.array(a)
return (s, q, a)
def pad_sentences(sentences,
sentence_length=None,
dtype=np.int32,
pad_val=0.):
logger.error(
'pad_sentances in the Text class is deprecated. This function'
'is now in neon.data.text_preprocessing.')
return pad_sentences(sentences,
sentence_length=sentence_length,
dtype=dtype,
pad_val=pad_val)
def vectorize_stories(self, data):
"""
Convert (story, query, answer) word data into vectors.
Args:
data (tuple) : Tuple of story, query, answer word data.
Returns:
tuple : Tuple of story, query, answer vectors.
"""
s, q, a = [], [], []
for story, query, answer in data:
s.append(self.words_to_vector(story))
q.append(self.words_to_vector(query))
a.append(self.one_hot_vector(answer))
s = pad_sentences(s, self.story_maxlen)
q = pad_sentences(q, self.query_maxlen)
a = np.array(a)
return (s, q, a)
def vectorize(words, max_len):
return be.array(pad_sentences([babi.words_to_vector(BABI.tokenize(words))], max_len))
def vectorize(words, max_len):
return be.array(
pad_sentences([babi.words_to_vector(BABI.tokenize(words))], max_len))
