def test():
_, _, _, sentence_size, vocab_size = build_corpus()
v2i, _ = build_vocab()
_, i2l = build_label()
origin_questions = ['今天 天气 不错', '介绍 贵金属 产品']
questions = [q.split() for q in origin_questions]
questions = [[v2i[vocab] for vocab in ques if vocab in v2i]
for ques in questions]
config = tf.ConfigProto()
with tf.Session(config=config) as sess:
model = Model(sentence_size, vocab_size, FLAGS.embed_size,
FLAGS.class_num, FLAGS.learning_rate, FLAGS.decay_step,
FLAGS.decay_rate, FLAGS.layer_size,
FLAGS.multi_channel_size)
saver = tf.train.Saver()
saver.restore(sess, tf.train.latest_checkpoint(FLAGS.check_point))
questions = pad_sequences(questions, maxlen=sentence_size, value=0)
feed_dict = {
model.encoder_input: questions,
model.batch_size: FLAGS.batch_size
}
p = sess.run([model.predict], feed_dict=feed_dict)
p = p[0].tolist()
for index in range(len(questions)):
print(f'{origin_questions[index]} is_business: {i2l[p[index]]}')
def main(_):
train, test, _, sentence_size, vocab_size = build_corpus()
train_x, train_y = train
test_x, test_y = test
config = tf.ConfigProto()
config.gpu_options.allow_growth = True
with tf.Session(config=config) as sess:
model = Model(sentence_size, FLAGS.class_num, vocab_size,
FLAGS.embed_size, FLAGS.filters, FLAGS.filter_num,
FLAGS.channel_size, FLAGS.keep_prob, FLAGS.learning_rate,
FLAGS.decay_step, FLAGS.decay_rate, FLAGS.k1,
FLAGS.k_top)
saver = tf.train.Saver()
if os.path.exists(FLAGS.check_point + 'checkpoint'):
saver.restore(sess, tf.train.latest_checkpoint(FLAGS.checkpoint))
else:
sess.run(tf.global_variables_initializer())
cur_epoch = sess.run(model.epoch_step)
corpus_size = len(train_x)
for epoch in range(cur_epoch, FLAGS.epoch_num):
loss, acc, count = 0, 0, 1
for start, end in zip(
range(0, corpus_size, FLAGS.batch_size),
range(FLAGS.batch_size, corpus_size, FLAGS.batch_size)):
_loss, _, _, _acc = sess.run([
model.loss, model.optimize, model.predict, model.accuracy
],
feed_dict={
model.x: train_x[start:end],
model.y: train_y[start:end]
})
loss, acc, count = loss + _loss, acc + _acc, count + 1
if count % FLAGS.batch_size == 0:
print(
f'Train -- epoch: {epoch}, count: {count}, loss: {loss/count}, acc: {acc/count}'
)
sess.run(model.epoch_increment)
if epoch % FLAGS.epoch_val == 0:
cur_loss, cur_predict, cur_acc = sess.run(
[model.loss, model.predict, model.accuracy],
feed_dict={
model.x: test_x,
model.y: test_y
})
print(
f'Eval -- epoch: {epoch}, loss: {cur_loss}, acc: {cur_acc}'
)
save_path = FLAGS.check_point + "a04_dcnn.model"
saver.save(sess, save_path, global_step=model.epoch_step)
def main(_):
train, test, _, sentence_size, vocab_size = build_corpus()
train_x, train_y = train
test_x, test_y = test
config = tf.ConfigProto()
config.gpu_options.allow_growth = True
with tf.Session(config=config) as sess:
model = Model(sentence_size, vocab_size, FLAGS.embed_size,
FLAGS.hidden_size, FLAGS.class_num, FLAGS.learning_rate,
FLAGS.decay_step, FLAGS.decay_rate, FLAGS.mode)
saver = tf.train.Saver()
if os.path.exists(FLAGS.check_point + 'checkpoint'):
saver.restore(sess, tf.train.latest_checkpoint(FLAGS.checkpoint))
else:
sess.run(tf.global_variables_initializer())
cur_epoch = sess.run(model.epoch_step)
corpus_size = len(train_x)
for epoch in range(cur_epoch, FLAGS.epoch_num):
loss, acc, count = 0, 0, 1
for start, end in zip(
range(0, corpus_size, FLAGS.batch_size),
range(FLAGS.batch_size, corpus_size, FLAGS.batch_size)):
train_yy = np.array([[2, i] + [i + 1 if i == 0 else i - 1]
for i in train_y[start:end]],
dtype=np.int32)
_loss, _, _, _acc = sess.run(
[
model.loss, model.optimize, model.predict,
model.accuracy
],
feed_dict={
model.encoder_input: train_x[start:end],
model.decoder_input: train_yy,
model.decoder_output: train_y[start:end],
model.batch_size: FLAGS.batch_size
})
loss, acc, count = loss + _loss, acc + _acc, count + 1
if count % FLAGS.batch_size == 0:
print(
f'Train -- epoch: {epoch}, count: {count}, loss: {loss/count}, acc: {acc/count}'
)
sess.run(model.epoch_increment)
if epoch % FLAGS.epoch_val == 0:
test_yy = np.array([[2, i] + [i + 1 if i == 0 else i - 1]
for i in test_y],
dtype=np.int32)
cur_loss, cur_predict, cur_acc = sess.run(
[model.loss, model.predict, model.accuracy],
feed_dict={
model.encoder_input: test_x,
model.decoder_input: test_yy,
model.decoder_output: test_y,
model.batch_size: len(test_x)
})
print(
f'Eval -- epoch: {epoch}, loss: {cur_loss}, acc: {cur_acc}'
)
save_path = FLAGS.check_point + "a05_attention.model"
saver.save(sess, save_path, global_step=model.epoch_step)