def get_validation_data(input_texts, target_texts, char2id, maxlen=400):
# 数据生成器
X, Y = [], []
for i in range(len(input_texts)):
X.append(str2id(input_texts[i], char2id, maxlen))
Y.append(str2id(target_texts[i], char2id, maxlen))
X = np.array(padding(X, char2id))
Y = np.array(padding(Y, char2id))
return [X, Y], None
def data_generator(input_texts, target_texts, char2id, batch_size, maxlen=400):
# 数据生成器
while True:
X, Y = [], []
for i in range(len(input_texts)):
X.append(str2id(input_texts[i], char2id, maxlen))
Y.append(str2id(target_texts[i], char2id, maxlen))
if len(X) == batch_size:
X = np.array(padding(X, char2id))
Y = np.array(padding(Y, char2id))
yield [X, Y], None
X, Y = [], []
def get_validation_data(input_texts,
target_texts,
char2id,
input_pinyins,
output_pinyins,
pingyin2id,
maxlen=400):
# 数据生成器
X, Y = [], []
X_p, Y_p = [], []
for i in range(len(input_texts)):
X.append(str2id(input_texts[i], char2id, maxlen))
Y.append(str2id(target_texts[i], char2id, maxlen))
X_p.append(str2id(input_pinyins[i], pingyin2id, maxlen))
Y_p.append(str2id(output_pinyins[i], pingyin2id, maxlen))
X = np.array(padding(X, char2id))
Y = np.array(padding(Y, char2id))
X_p = np.array(padding(X_p, pingyin2id))
Y_p = np.array(padding(Y_p, pingyin2id))
return [X, Y, X_p, Y_p], None
def data_generator(input_texts,
target_texts,
char2id,
input_pinyins,
output_pinyins,
pingyin2id,
batch_size,
maxlen=400):
# 数据生成器
while True:
X, Y = [], []
X_p, Y_p = [], []
for i in range(len(input_texts)):
X.append(str2id(input_texts[i], char2id, maxlen))
Y.append(str2id(target_texts[i], char2id, maxlen))
X_p.append(str2id(input_pinyins[i], pingyin2id, maxlen))
Y_p.append(str2id(output_pinyins[i], pingyin2id, maxlen))
if len(X) == batch_size:
X = np.array(padding(X, char2id))
Y = np.array(padding(Y, char2id))
X_p = np.array(padding(X_p, pingyin2id))
Y_p = np.array(padding(Y_p, pingyin2id))
yield [X, Y, X_p, Y_p], None
X, Y, X_p, Y_p = [], [], [], []
