def __init__(self, origin_path=get_current_path('data/rnn.csv'), train_path=get_current_path('data/rnn_train.csv'),
model_path=get_current_path('models/rnn')):
super(TrainRNN, self).__init__(origin_path, train_path, model_path)
# Load "Penn Treebank" dataset
self.corpus = Corpus()
self.ids = self.corpus.get_data(self.train_path, batch_size)
self.vocab_size = len(self.corpus.dictionary)
self.num_batches = self.ids.size(1) // seq_length
self.dict_path = get_current_path('models/rnn_dict')
def __init__(self, search_list, *args):
"""
Keyword arguments:
rnnlm -- 语言模型，RNN+LSTM
question -- 候选生成问句
"""
self.rnnlm = torch.load(get_current_path(args[0]))
with open(get_current_path(args[1]), mode='rb') as f:
self.dict = pickle.load(f)
self.s1, self.s2, self.s3 = '', '', ''
for index, i in enumerate(search_list):
if index == 0:
self.s1 = i
elif index == 1:
self.s2 = i
elif index == 2:
self.s3 = i
self.questions = searchs2templates(s1=self.s1,
s2=self.s2,
s3=self.s3)
def __init__(self, search_list, svm_path):
"""
Keyword arguments:
clf -- 分类器，svm
x1 -- x1，查询词1
x2 -- x2，查询词2
x3 -- x3,查询词3
"""
self.x1, self.x2, self.x3 = '', '', ''
self.clf = joblib.load(get_current_path(svm_path))
for index, i in enumerate(search_list):
if index == 0:
self.x1 = i
elif index == 1:
self.x2 = i
elif index == 2:
self.x3 = i
def __init__(self, search_list, w2v_path, random):
"""
Keyword arguments:
w2v -- 词嵌入，word2vec
templates -- 搜索序列和模板号
"""
self.x1, self.x2, self.x3 = '', '', ''
self.w2v = Word2Vec.load(get_current_path(w2v_path))
for index, i in enumerate(search_list):
if index == 0:
self.x1 = i
elif index == 1:
self.x2 = i
elif index == 2:
self.x3 = i
self.templates = templates(x1=self.x1, x2=self.x2, x3=self.x3)
self.random = random
def __init__(self, origin_path=get_current_path('data/svm.csv'), train_path=get_current_path('data/svm_train.csv'),
model_path=get_current_path('models/svm')):
super(TrainSVM, self).__init__(origin_path, train_path, model_path)
def __init__(self, origin_path=get_current_path('data/w2v.csv'), train_path=get_current_path('data/w2v_train.csv'),
model_path=get_current_path('models/w2v')):
super(TrainW2V, self).__init__(origin_path, train_path, model_path)